diff --git a/README.md b/README.md index 6d04f0c9e..277f25fb8 100644 --- a/README.md +++ b/README.md @@ -2,7 +2,7 @@ *Visual instruction tuning towards large language and vision models with GPT-4 level capabilities.* -[[Project Page](https://llava-vl.github.io/)] [[Demo](https://llava.hliu.cc/)] [[Data](https://github.com/haotian-liu/LLaVA/blob/main/docs/Data.md)] [[Model Zoo](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md)] +[📢 [LLaVA-1.6 Blog](https://llava-vl.github.io/blog/2024-01-30-llava-1-6/)] [[Project Page](https://llava-vl.github.io/)] [[Demo](https://llava.hliu.cc/)] [[Data](https://github.com/haotian-liu/LLaVA/blob/main/docs/Data.md)] [[Model Zoo](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md)] 🤝Community Contributions: [[llama.cpp](https://github.com/ggerganov/llama.cpp/pull/3436)] [[Colab](https://github.com/camenduru/LLaVA-colab)] [[🤗Space](https://huggingface.co/spaces/badayvedat/LLaVA)] [[Replicate](https://replicate.com/yorickvp/llava-13b)] [[AutoGen](https://github.com/microsoft/autogen/blob/main/notebook/agentchat_lmm_llava.ipynb)] [[BakLLaVA](https://github.com/SkunkworksAI/BakLLaVA)] @@ -19,6 +19,7 @@ ## Release +- [1/30] 🔥 LLaVA-1.6 is out! With additional scaling to LLaVA-1.5, LLaVA-1.6-34B outperforms Gemini Pro. It can now process 4x more pixels and perform more tasks/applications than before. Check out the [blog post](https://llava-vl.github.io/blog/2024-01-30-llava-1-6/), and explore the [demo](https://llava.hliu.cc/)! Models are available in [Model Zoo](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md). Training/eval data and scripts coming soon. - [11/10] [LLaVA-Plus](https://llava-vl.github.io/llava-plus/) is released: Learning to Use Tools for Creating Multimodal Agents, with LLaVA-Plus (LLaVA that Plug and Learn to Use Skills). [[Project Page](https://llava-vl.github.io/llava-plus/)] [[Demo](https://llavaplus.ngrok.io/)] [[Code](https://github.com/LLaVA-VL/LLaVA-Plus-Codebase)] [[Paper](https://arxiv.org/abs/2311.05437)] - [11/2] [LLaVA-Interactive](https://llava-vl.github.io/llava-interactive/) is released: Experience the future of human-AI multimodal interaction with an all-in-one demo for Image Chat, Segmentation, Generation and Editing. [[Project Page](https://llava-vl.github.io/llava-interactive/)] [[Demo](https://llavainteractive.ngrok.io/)] [[Code](https://github.com/LLaVA-VL/LLaVA-Interactive-Demo)] [[Paper](https://arxiv.org/abs/2311.00571)] - [10/26] 🔥 LLaVA-1.5 with LoRA achieves comparable performance as full-model finetuning, with a reduced GPU RAM requirement ([ckpts](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md#llava-v15), [script](https://github.com/haotian-liu/LLaVA#train)). We also provide a [doc](https://github.com/haotian-liu/LLaVA/blob/main/docs/Finetune_Custom_Data.md) on how to finetune LLaVA-1.5 on your own dataset with LoRA. diff --git a/docs/MODEL_ZOO.md b/docs/MODEL_ZOO.md index 2ab48d58a..577218226 100644 --- a/docs/MODEL_ZOO.md +++ b/docs/MODEL_ZOO.md @@ -4,7 +4,19 @@ If you are interested in including any other details in Model Zoo, please open an issue :) -The model weights below are *merged* weights. You do not need to apply delta. The usage of LLaVA checkpoints should comply with the base LLM's model license: [Llama 2](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md). +The model weights below are *merged* weights. You do not need to apply delta. The usage of LLaVA checkpoints should comply with the base LLM's model license. + +## LLaVA-v1.6 + +| Version | LLM | Schedule | Checkpoint | MMMU | MathVista | VQAv2 | GQA | VizWiz | SQA | TextVQA | POPE | MME | MM-Bench | MM-Bench-CN | SEED-IMG | LLaVA-Bench-Wild | MM-Vet | +|----------|----------|-----------|-----------|---|---|---|---|---|---|---|---|---|---|---|---|---|---| +| LLaVA-1.6 | Vicuna-7B | full_ft-1e | [liuhaotian/llava-v1.6-vicuna-7b](https://huggingface.co/liuhaotian/llava-v1.6-vicuna-7b) | 35.8 | 34.6 | 81.8 | 64.2 | 57.6 | 70.1 | 64.9 | 86.5 | 1519/332 | 67.4 | 60.6 | 70.2 | 81.6 | 43.9 | +| LLaVA-1.6 | Vicuna-13B | full_ft-1e | [liuhaotian/llava-v1.6-vicuna-13b](https://huggingface.co/liuhaotian/llava-v1.6-vicuna-13b) | 36.2 | 35.3 | 82.8 | 65.4 | 60.5 | 73.6 | 67.1 | 86.2 | 1575/326 | 70 | 64.4 | 71.9 | 87.3 | 48.4 | +| LLaVA-1.6 | Mistral-7B | full_ft-1e | [liuhaotian/llava-v1.6-mistral-7b](https://huggingface.co/liuhaotian/llava-v1.6-mistral-7b) | 35.3 | 37.7 | 82.2 | 64.8 | 60.0 | 72.8 | 65.7 | 86.7 | 1498/321 | 68.7 | 61.2 | 72.2 | 83.2 | 47.3 | +| LLaVA-1.6 | Hermes-Yi-34B | full_ft-1e | [liuhaotian/llava-v1.6-34b](https://huggingface.co/liuhaotian/llava-v1.6-34b) | 51.1 | 46.5 | 83.7 | 67.1 | 63.8 | 81.8 | 69.5 | 87.7 | 1631/397 | 79.3 | 79 | 75.9 | 89.6 | 57.4 | + +*LLaVA-1.6-34B outperforms Gemini Pro on benchmarks like MMMU and MathVista.* + ## LLaVA-v1.5 diff --git a/llava/conversation.py b/llava/conversation.py index 0025f5b11..65ee05d5e 100644 --- a/llava/conversation.py +++ b/llava/conversation.py @@ -68,7 +68,7 @@ def get_prompt(self): else: ret += role elif self.sep_style == SeparatorStyle.LLAMA_2: - wrap_sys = lambda msg: f"<>\n{msg}\n<>\n\n" + wrap_sys = lambda msg: f"<>\n{msg}\n<>\n\n" if len(msg) > 0 else msg wrap_inst = lambda msg: f"[INST] {msg} [/INST]" ret = "" @@ -357,6 +357,28 @@ def dict(self): version="v1_mmtag", ) +conv_mistral_instruct = Conversation( + system="", + roles=("USER", "ASSISTANT"), + version="llama_v2", + messages=(), + offset=0, + sep_style=SeparatorStyle.LLAMA_2, + sep="", + sep2="", +) + +conv_chatml_direct = Conversation( + system="""<|im_start|>system +Answer the questions.""", + roles=("<|im_start|>user\n", "<|im_start|>assistant\n"), + version="mpt", + messages=(), + offset=0, + sep_style=SeparatorStyle.MPT, + sep="<|im_end|>", +) + default_conversation = conv_vicuna_v1 conv_templates = { "default": conv_vicuna_v0, @@ -364,6 +386,9 @@ def dict(self): "v1": conv_vicuna_v1, "vicuna_v1": conv_vicuna_v1, "llama_2": conv_llama_2, + "mistral_instruct": conv_mistral_instruct, + "chatml_direct": conv_chatml_direct, + "mistral_direct": conv_chatml_direct, "plain": conv_llava_plain, "v0_plain": conv_llava_plain, diff --git a/llava/eval/model_vqa.py b/llava/eval/model_vqa.py index 59dca734c..938706438 100644 --- a/llava/eval/model_vqa.py +++ b/llava/eval/model_vqa.py @@ -9,7 +9,7 @@ from llava.conversation import conv_templates, SeparatorStyle from llava.model.builder import load_pretrained_model from llava.utils import disable_torch_init -from llava.mm_utils import tokenizer_image_token, get_model_name_from_path, KeywordsStoppingCriteria +from llava.mm_utils import tokenizer_image_token, process_images, get_model_name_from_path from PIL import Image import math @@ -55,17 +55,14 @@ def eval_model(args): input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() - image = Image.open(os.path.join(args.image_folder, image_file)) - image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0] - - stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 - keywords = [stop_str] - stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids) + image = Image.open(os.path.join(args.image_folder, image_file)).convert('RGB') + image_tensor = process_images([image], image_processor, model.config)[0] with torch.inference_mode(): output_ids = model.generate( input_ids, images=image_tensor.unsqueeze(0).half().cuda(), + image_sizes=[image.size], do_sample=True if args.temperature > 0 else False, temperature=args.temperature, top_p=args.top_p, @@ -74,15 +71,7 @@ def eval_model(args): max_new_tokens=1024, use_cache=True) - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() - if outputs.endswith(stop_str): - outputs = outputs[:-len(stop_str)] - outputs = outputs.strip() + outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ans_id = shortuuid.uuid() ans_file.write(json.dumps({"question_id": idx, diff --git a/llava/eval/model_vqa_loader.py b/llava/eval/model_vqa_loader.py index a50944db6..d435b7d83 100644 --- a/llava/eval/model_vqa_loader.py +++ b/llava/eval/model_vqa_loader.py @@ -55,17 +55,24 @@ def __getitem__(self, index): input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt') - return input_ids, image_tensor + return input_ids, image_tensor, image.size def __len__(self): return len(self.questions) +def collate_fn(batch): + input_ids, image_tensors, image_sizes = zip(*batch) + input_ids = torch.stack(input_ids, dim=0) + image_tensors = torch.stack(image_tensors, dim=0) + return input_ids, image_tensors, image_sizes + + # DataLoader def create_data_loader(questions, image_folder, tokenizer, image_processor, model_config, batch_size=1, num_workers=4): assert batch_size == 1, "batch_size must be 1" dataset = CustomDataset(questions, image_folder, tokenizer, image_processor, model_config) - data_loader = DataLoader(dataset, batch_size=batch_size, num_workers=num_workers, shuffle=False) + data_loader = DataLoader(dataset, batch_size=batch_size, num_workers=num_workers, shuffle=False, collate_fn=collate_fn) return data_loader @@ -88,7 +95,7 @@ def eval_model(args): data_loader = create_data_loader(questions, args.image_folder, tokenizer, image_processor, model.config) - for (input_ids, image_tensor), line in tqdm(zip(data_loader, questions), total=len(questions)): + for (input_ids, image_tensor, image_sizes), line in tqdm(zip(data_loader, questions), total=len(questions)): idx = line["question_id"] cur_prompt = line["text"] @@ -98,6 +105,7 @@ def eval_model(args): output_ids = model.generate( input_ids, images=image_tensor.to(dtype=torch.float16, device='cuda', non_blocking=True), + image_sizes=image_sizes, do_sample=True if args.temperature > 0 else False, temperature=args.temperature, top_p=args.top_p, @@ -105,12 +113,7 @@ def eval_model(args): max_new_tokens=args.max_new_tokens, use_cache=True) - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() + outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ans_id = shortuuid.uuid() ans_file.write(json.dumps({"question_id": idx, diff --git a/llava/eval/model_vqa_mmbench.py b/llava/eval/model_vqa_mmbench.py index 2ffec1b59..bd7a4c808 100644 --- a/llava/eval/model_vqa_mmbench.py +++ b/llava/eval/model_vqa_mmbench.py @@ -106,14 +106,12 @@ def eval_model(args): input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() image_tensor = process_images([image], image_processor, model.config)[0] - # image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0] - - stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 with torch.inference_mode(): output_ids = model.generate( input_ids, images=image_tensor.unsqueeze(0).half().cuda(), + image_sizes=[image.size], do_sample=True if args.temperature > 0 else False, temperature=args.temperature, top_p=args.top_p, @@ -122,15 +120,7 @@ def eval_model(args): max_new_tokens=1024, use_cache=True) - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() - if outputs.endswith(stop_str): - outputs = outputs[:-len(stop_str)] - outputs = outputs.strip() + outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ans_id = shortuuid.uuid() ans_file.write(json.dumps({"question_id": idx, diff --git a/llava/eval/model_vqa_qbench.py b/llava/eval/model_vqa_qbench.py deleted file mode 100644 index f3ca8177c..000000000 --- a/llava/eval/model_vqa_qbench.py +++ /dev/null @@ -1,122 +0,0 @@ -import argparse -import torch -from tqdm import tqdm -import json - -from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN -from llava.conversation import conv_templates, SeparatorStyle -from llava.model.builder import load_pretrained_model -from llava.utils import disable_torch_init -from llava.mm_utils import tokenizer_image_token, get_model_name_from_path, KeywordsStoppingCriteria - -from PIL import Image - -import requests -from PIL import Image -from io import BytesIO - - -def load_image(image_file): - if image_file.startswith('http') or image_file.startswith('https'): - response = requests.get(image_file) - image = Image.open(BytesIO(response.content)).convert('RGB') - else: - image = Image.open(image_file).convert('RGB') - return image - - -def eval_model(args): - # Model - disable_torch_init() - - model_name = get_model_name_from_path(args.model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(args.model_path, args.model_base, model_name, True) - - - - - with open(args.questions_file) as f: - llvqa_data = json.load(f) - - for i, llddata in enumerate(tqdm(llvqa_data)): - filename = llddata["img_path"] - if args.lang == "en": - message = llddata["question"] + "\nChoose between one of the options as follows:\n" - elif args.lang == "zh": - message = llddata["question"] + "\在下列选项中选择一个:\n" - else: - raise NotImplementedError("Q-Bench does not support languages other than English (en) and Chinese (zh) yet. Contact us (https://github.com/VQAssessment/Q-Bench/) to convert Q-Bench into more languages.") - for choice, ans in zip(["A.", "B.", "C.", "D."], llddata["candidates"]): - message += f"{choice} {ans}\n" - qs = message - - if model.config.mm_use_im_start_end: - qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs - else: - qs = DEFAULT_IMAGE_TOKEN + '\n' + qs - - if 'llama-2' in model_name.lower(): - conv_mode = "llava_llama_2" - elif "v1" in model_name.lower(): - conv_mode = "llava_v1" - elif "mpt" in model_name.lower(): - conv_mode = "mpt" - else: - conv_mode = "llava_v0" - - if args.conv_mode is not None and conv_mode != args.conv_mode: - print('[WARNING] the auto inferred conversation mode is {}, while `--conv-mode` is {}, using {}'.format(conv_mode, args.conv_mode, args.conv_mode)) - else: - args.conv_mode = conv_mode - - conv = conv_templates[args.conv_mode].copy() - conv.append_message(conv.roles[0], qs) - conv.append_message(conv.roles[1], None) - prompt = conv.get_prompt() - - image = load_image(args.image_folder + filename) - image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'].half().cuda() - - input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() - - stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 - keywords = [stop_str] - stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids) - - - with torch.inference_mode(): - output_ids = model.generate( - input_ids, - images=image_tensor, - num_beams=1, - do_sample=False, - temperature=0, - max_new_tokens=1024, - use_cache=True, - stopping_criteria=[stopping_criteria]) - - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() - if outputs.endswith(stop_str): - outputs = outputs[:-len(stop_str)] - outputs = outputs.strip() - llddata["response"] = outputs - with open(args.answers_file, "a") as wf: - json.dump(llddata, wf) - -if __name__ == "__main__": - parser = argparse.ArgumentParser() - parser.add_argument("--model-path", type=str, default="llava-v1.5") - parser.add_argument("--model-base", type=str, default=None) - parser.add_argument("--image-folder", type=str, default="./playground/data/qbench/images_llvisionqa") - parser.add_argument("--questions-file", type=str, default="./playground/data/qbench/llvisionqa_dev.json") - parser.add_argument("--answers-file", type=str, default="answer.jsonl") - parser.add_argument("--conv-mode", type=str, default="llava_v1") - parser.add_argument("--lang", type=str, default="en") - args = parser.parse_args() - - eval_model(args) diff --git a/llava/eval/model_vqa_science.py b/llava/eval/model_vqa_science.py index e99501f2b..90fc681a2 100644 --- a/llava/eval/model_vqa_science.py +++ b/llava/eval/model_vqa_science.py @@ -9,7 +9,7 @@ from llava.conversation import conv_templates, SeparatorStyle from llava.model.builder import load_pretrained_model from llava.utils import disable_torch_init -from llava.mm_utils import tokenizer_image_token, get_model_name_from_path, KeywordsStoppingCriteria +from llava.mm_utils import tokenizer_image_token, process_images, get_model_name_from_path from PIL import Image import math @@ -47,8 +47,9 @@ def eval_model(args): if 'image' in line: image_file = line["image"] image = Image.open(os.path.join(args.image_folder, image_file)) - image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0] + image_tensor = process_images([image], image_processor, model.config)[0] images = image_tensor.unsqueeze(0).half().cuda() + image_sizes = [image.size] if getattr(model.config, 'mm_use_im_start_end', False): qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs else: @@ -56,6 +57,7 @@ def eval_model(args): cur_prompt = '' + '\n' + cur_prompt else: images = None + image_sizes = None if args.single_pred_prompt: qs = qs + '\n' + "Answer with the option's letter from the given choices directly." @@ -68,56 +70,18 @@ def eval_model(args): input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() - stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 - keywords = [stop_str] - stopping_criteria = [KeywordsStoppingCriteria(keywords, tokenizer, input_ids)] if conv.version == "v0" else None - with torch.inference_mode(): output_ids = model.generate( input_ids, images=images, + image_sizes=image_sizes, do_sample=True if args.temperature > 0 else False, temperature=args.temperature, max_new_tokens=1024, use_cache=True, - stopping_criteria=stopping_criteria, ) - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() - if outputs.endswith(stop_str): - outputs = outputs[:-len(stop_str)] - outputs = outputs.strip() - - # prompt for answer - if args.answer_prompter: - outputs_reasoning = outputs - input_ids = tokenizer_image_token(prompt + outputs_reasoning + ' ###\nANSWER:', tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() - - with torch.inference_mode(): - output_ids = model.generate( - input_ids, - images=images, - do_sample=True if args.temperature > 0 else False, - temperature=args.temperature, - max_new_tokens=64, - use_cache=True, - stopping_criteria=[stopping_criteria]) - - input_token_len = input_ids.shape[1] - n_diff_input_output = (input_ids != output_ids[:, :input_token_len]).sum().item() - if n_diff_input_output > 0: - print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids') - outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)[0] - outputs = outputs.strip() - if outputs.endswith(stop_str): - outputs = outputs[:-len(stop_str)] - outputs = outputs.strip() - outputs = outputs_reasoning + '\n The answer is ' + outputs + outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ans_id = shortuuid.uuid() ans_file.write(json.dumps({"question_id": idx, diff --git a/llava/mm_utils.py b/llava/mm_utils.py index c4b8862ff..71092ff1d 100644 --- a/llava/mm_utils.py +++ b/llava/mm_utils.py @@ -1,12 +1,150 @@ from PIL import Image from io import BytesIO import base64 - import torch +import math +import ast + from transformers import StoppingCriteria from llava.constants import IMAGE_TOKEN_INDEX +def select_best_resolution(original_size, possible_resolutions): + """ + Selects the best resolution from a list of possible resolutions based on the original size. + + Args: + original_size (tuple): The original size of the image in the format (width, height). + possible_resolutions (list): A list of possible resolutions in the format [(width1, height1), (width2, height2), ...]. + + Returns: + tuple: The best fit resolution in the format (width, height). + """ + original_width, original_height = original_size + best_fit = None + max_effective_resolution = 0 + min_wasted_resolution = float('inf') + + for width, height in possible_resolutions: + scale = min(width / original_width, height / original_height) + downscaled_width, downscaled_height = int(original_width * scale), int(original_height * scale) + effective_resolution = min(downscaled_width * downscaled_height, original_width * original_height) + wasted_resolution = (width * height) - effective_resolution + + if effective_resolution > max_effective_resolution or (effective_resolution == max_effective_resolution and wasted_resolution < min_wasted_resolution): + max_effective_resolution = effective_resolution + min_wasted_resolution = wasted_resolution + best_fit = (width, height) + + return best_fit + + +def resize_and_pad_image(image, target_resolution): + """ + Resize and pad an image to a target resolution while maintaining aspect ratio. + + Args: + image (PIL.Image.Image): The input image. + target_resolution (tuple): The target resolution (width, height) of the image. + + Returns: + PIL.Image.Image: The resized and padded image. + """ + original_width, original_height = image.size + target_width, target_height = target_resolution + + scale_w = target_width / original_width + scale_h = target_height / original_height + + if scale_w < scale_h: + new_width = target_width + new_height = min(math.ceil(original_height * scale_w), target_height) + else: + new_height = target_height + new_width = min(math.ceil(original_width * scale_h), target_width) + + # Resize the image + resized_image = image.resize((new_width, new_height)) + + new_image = Image.new('RGB', (target_width, target_height), (0, 0, 0)) + paste_x = (target_width - new_width) // 2 + paste_y = (target_height - new_height) // 2 + new_image.paste(resized_image, (paste_x, paste_y)) + + return new_image + + +def divide_to_patches(image, patch_size): + """ + Divides an image into patches of a specified size. + + Args: + image (PIL.Image.Image): The input image. + patch_size (int): The size of each patch. + + Returns: + list: A list of PIL.Image.Image objects representing the patches. + """ + patches = [] + width, height = image.size + for i in range(0, height, patch_size): + for j in range(0, width, patch_size): + box = (j, i, j + patch_size, i + patch_size) + patch = image.crop(box) + patches.append(patch) + + return patches + + +def get_anyres_image_grid_shape(image_size, grid_pinpoints, patch_size): + """ + Calculate the shape of the image patch grid after the preprocessing for images of any resolution. + + Args: + image_size (tuple): The size of the input image in the format (width, height). + grid_pinpoints (str): A string representation of a list of possible resolutions. + patch_size (int): The size of each image patch. + + Returns: + tuple: The shape of the image patch grid in the format (width, height). + """ + if type(grid_pinpoints) is list: + possible_resolutions = grid_pinpoints + else: + possible_resolutions = ast.literal_eval(grid_pinpoints) + width, height = select_best_resolution(image_size, possible_resolutions) + return width // patch_size, height // patch_size + + +def process_anyres_image(image, processor, grid_pinpoints): + """ + Process an image with variable resolutions. + + Args: + image (PIL.Image.Image): The input image to be processed. + processor: The image processor object. + grid_pinpoints (str): A string representation of a list of possible resolutions. + + Returns: + torch.Tensor: A tensor containing the processed image patches. + """ + if type(grid_pinpoints) is list: + possible_resolutions = grid_pinpoints + else: + possible_resolutions = ast.literal_eval(grid_pinpoints) + best_resolution = select_best_resolution(image.size, possible_resolutions) + image_padded = resize_and_pad_image(image, best_resolution) + + patches = divide_to_patches(image_padded, processor.crop_size['height']) + + image_original_resize = image.resize((processor.size['shortest_edge'], processor.size['shortest_edge'])) + + image_patches = [image_original_resize] + patches + image_patches = [processor.preprocess(image_patch, return_tensors='pt')['pixel_values'][0] + for image_patch in image_patches] + return torch.stack(image_patches, dim=0) + + def load_image_from_base64(image): return Image.open(BytesIO(base64.b64decode(image))) @@ -33,6 +171,10 @@ def process_images(images, image_processor, model_cfg): image = expand2square(image, tuple(int(x*255) for x in image_processor.image_mean)) image = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0] new_images.append(image) + elif image_aspect_ratio == "anyres": + for image in images: + image = process_anyres_image(image, image_processor, model_cfg.image_grid_pinpoints) + new_images.append(image) else: return image_processor(images, return_tensors='pt')['pixel_values'] if all(x.shape == new_images[0].shape for x in new_images): diff --git a/llava/model/__init__.py b/llava/model/__init__.py index fa7996054..dbd91789f 100644 --- a/llava/model/__init__.py +++ b/llava/model/__init__.py @@ -1,2 +1,6 @@ -from .language_model.llava_llama import LlavaLlamaForCausalLM, LlavaConfig -from .language_model.llava_mpt import LlavaMPTForCausalLM, LlavaMPTConfig +try: + from .language_model.llava_llama import LlavaLlamaForCausalLM, LlavaConfig + from .language_model.llava_mpt import LlavaMptForCausalLM, LlavaMptConfig + from .language_model.llava_mistral import LlavaMistralForCausalLM, LlavaMistralConfig +except: + pass diff --git a/llava/model/builder.py b/llava/model/builder.py index ca9c81a7a..33edbd100 100644 --- a/llava/model/builder.py +++ b/llava/model/builder.py @@ -101,6 +101,14 @@ def load_from_hf(repo_id, filename, subfolder=None): if 'mpt' in model_name.lower(): tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True) model = LlavaMPTForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) + elif 'mistral' in model_name.lower(): + tokenizer = AutoTokenizer.from_pretrained(model_path) + model = LlavaMistralForCausalLM.from_pretrained( + model_path, + low_cpu_mem_usage=True, + use_flash_attention_2=False, + **kwargs + ) else: tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False) model = LlavaLlamaForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) diff --git a/llava/model/language_model/llava_llama.py b/llava/model/language_model/llava_llama.py index 58ccb3057..069d0d1c1 100644 --- a/llava/model/language_model/llava_llama.py +++ b/llava/model/language_model/llava_llama.py @@ -22,12 +22,13 @@ LlamaConfig, LlamaModel, LlamaForCausalLM from transformers.modeling_outputs import CausalLMOutputWithPast +from transformers.generation.utils import GenerateOutput from ..llava_arch import LlavaMetaModel, LlavaMetaForCausalLM class LlavaConfig(LlamaConfig): - model_type = "llava" + model_type = "llava_llama" class LlavaLlamaModel(LlavaMetaModel, LlamaModel): @@ -65,6 +66,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, images: Optional[torch.FloatTensor] = None, + image_sizes: Optional[List[List[int]]] = None, return_dict: Optional[bool] = None, ) -> Union[Tuple, CausalLMOutputWithPast]: @@ -82,7 +84,8 @@ def forward( attention_mask, past_key_values, labels, - images + images, + image_sizes ) return super().forward( @@ -98,14 +101,58 @@ def forward( return_dict=return_dict ) - def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs): + @torch.no_grad() + def generate( + self, + inputs: Optional[torch.Tensor] = None, + images: Optional[torch.Tensor] = None, + image_sizes: Optional[torch.Tensor] = None, + **kwargs, + ) -> Union[GenerateOutput, torch.LongTensor]: + position_ids = kwargs.pop("position_ids", None) + attention_mask = kwargs.pop("attention_mask", None) + if "inputs_embeds" in kwargs: + raise NotImplementedError("`inputs_embeds` is not supported") + + if images is not None: + ( + inputs, + position_ids, + attention_mask, + _, + inputs_embeds, + _ + ) = self.prepare_inputs_labels_for_multimodal( + inputs, + position_ids, + attention_mask, + None, + None, + images, + image_sizes=image_sizes + ) + else: + inputs_embeds = self.get_model().embed_tokens(inputs) + + return super().generate( + position_ids=position_ids, + attention_mask=attention_mask, + inputs_embeds=inputs_embeds, + **kwargs + ) + + def prepare_inputs_for_generation(self, input_ids, past_key_values=None, + inputs_embeds=None, **kwargs): images = kwargs.pop("images", None) - _inputs = super().prepare_inputs_for_generation( + image_sizes = kwargs.pop("image_sizes", None) + inputs = super().prepare_inputs_for_generation( input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs ) if images is not None: - _inputs['images'] = images - return _inputs + inputs['images'] = images + if image_sizes is not None: + inputs['image_sizes'] = image_sizes + return inputs -AutoConfig.register("llava", LlavaConfig) +AutoConfig.register("llava_llama", LlavaConfig) AutoModelForCausalLM.register(LlavaConfig, LlavaLlamaForCausalLM) diff --git a/llava/model/language_model/llava_mistral.py b/llava/model/language_model/llava_mistral.py new file mode 100644 index 000000000..0def682ea --- /dev/null +++ b/llava/model/language_model/llava_mistral.py @@ -0,0 +1,158 @@ +# Copyright 2023 Haotian Liu +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from typing import List, Optional, Tuple, Union + +import torch +import torch.nn as nn +from torch.nn import CrossEntropyLoss + +from transformers import AutoConfig, AutoModelForCausalLM, \ + MistralConfig, MistralModel, MistralForCausalLM + +from transformers.modeling_outputs import CausalLMOutputWithPast +from transformers.generation.utils import GenerateOutput + +from ..llava_arch import LlavaMetaModel, LlavaMetaForCausalLM + + +class LlavaMistralConfig(MistralConfig): + model_type = "llava_mistral" + + +class LlavaMistralModel(LlavaMetaModel, MistralModel): + config_class = LlavaMistralConfig + + def __init__(self, config: MistralConfig): + super(LlavaMistralModel, self).__init__(config) + + +class LlavaMistralForCausalLM(MistralForCausalLM, LlavaMetaForCausalLM): + config_class = LlavaMistralConfig + + def __init__(self, config): + super(MistralForCausalLM, self).__init__(config) + self.model = LlavaMistralModel(config) + + self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) + + # Initialize weights and apply final processing + self.post_init() + + def get_model(self): + return self.model + + def forward( + self, + input_ids: torch.LongTensor = None, + attention_mask: Optional[torch.Tensor] = None, + position_ids: Optional[torch.LongTensor] = None, + past_key_values: Optional[List[torch.FloatTensor]] = None, + inputs_embeds: Optional[torch.FloatTensor] = None, + labels: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, + output_attentions: Optional[bool] = None, + output_hidden_states: Optional[bool] = None, + images: Optional[torch.FloatTensor] = None, + image_sizes: Optional[List[List[int]]] = None, + return_dict: Optional[bool] = None, + ) -> Union[Tuple, CausalLMOutputWithPast]: + + if inputs_embeds is None: + ( + input_ids, + position_ids, + attention_mask, + past_key_values, + inputs_embeds, + labels + ) = self.prepare_inputs_labels_for_multimodal( + input_ids, + position_ids, + attention_mask, + past_key_values, + labels, + images, + image_sizes + ) + + return super().forward( + input_ids=input_ids, + attention_mask=attention_mask, + position_ids=position_ids, + past_key_values=past_key_values, + inputs_embeds=inputs_embeds, + labels=labels, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict + ) + + @torch.no_grad() + def generate( + self, + inputs: Optional[torch.Tensor] = None, + images: Optional[torch.Tensor] = None, + image_sizes: Optional[torch.Tensor] = None, + **kwargs, + ) -> Union[GenerateOutput, torch.LongTensor]: + position_ids = kwargs.pop("position_ids", None) + attention_mask = kwargs.pop("attention_mask", None) + if "inputs_embeds" in kwargs: + raise NotImplementedError("`inputs_embeds` is not supported") + + if images is not None: + ( + inputs, + position_ids, + attention_mask, + _, + inputs_embeds, + _ + ) = self.prepare_inputs_labels_for_multimodal( + inputs, + position_ids, + attention_mask, + None, + None, + images, + image_sizes=image_sizes + ) + else: + inputs_embeds = self.get_model().embed_tokens(inputs) + + return super().generate( + position_ids=position_ids, + attention_mask=attention_mask, + inputs_embeds=inputs_embeds, + **kwargs + ) + + def prepare_inputs_for_generation(self, input_ids, past_key_values=None, + inputs_embeds=None, **kwargs): + images = kwargs.pop("images", None) + image_sizes = kwargs.pop("image_sizes", None) + inputs = super().prepare_inputs_for_generation( + input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs + ) + if images is not None: + inputs['images'] = images + if image_sizes is not None: + inputs['image_sizes'] = image_sizes + return inputs + +AutoConfig.register("llava_mistral", LlavaMistralConfig) +AutoModelForCausalLM.register(LlavaMistralConfig, LlavaMistralForCausalLM) diff --git a/llava/model/language_model/llava_mpt.py b/llava/model/language_model/llava_mpt.py index 0200d1f8c..02e5237ec 100644 --- a/llava/model/language_model/llava_mpt.py +++ b/llava/model/language_model/llava_mpt.py @@ -13,101 +13,85 @@ # limitations under the License. -from typing import List, Optional, Tuple -import warnings +from typing import Optional, Tuple import torch -import torch.nn.functional as F -import math -from transformers import AutoConfig, AutoModelForCausalLM -from transformers.modeling_outputs import CausalLMOutputWithPast - -from .mpt.modeling_mpt import MPTConfig, MPTForCausalLM, MPTModel +from transformers import AutoConfig, AutoModelForCausalLM, \ + MptConfig, MptForCausalLM, MptModel from llava.model.llava_arch import LlavaMetaModel, LlavaMetaForCausalLM -class LlavaMPTConfig(MPTConfig): +class LlavaMptConfig(MptConfig): model_type = "llava_mpt" -class LlavaMPTModel(LlavaMetaModel, MPTModel): - config_class = LlavaMPTConfig +class LlavaMptModel(LlavaMetaModel, MptModel): + config_class = LlavaMptConfig - def __init__(self, config: MPTConfig): + def __init__(self, config: MptConfig): config.hidden_size = config.d_model - super(LlavaMPTModel, self).__init__(config) + super(LlavaMptModel, self).__init__(config) def embed_tokens(self, x): return self.wte(x) -class LlavaMPTForCausalLM(MPTForCausalLM, LlavaMetaForCausalLM): - config_class = LlavaMPTConfig +class LlavaMptForCausalLM(MptForCausalLM, LlavaMetaForCausalLM): + config_class = LlavaMptConfig supports_gradient_checkpointing = True def __init__(self, config): - super(MPTForCausalLM, self).__init__(config) - - if not config.tie_word_embeddings: - raise ValueError('MPTForCausalLM only supports tied word embeddings') - self.transformer = LlavaMPTModel(config) - self.logit_scale = None - if config.logit_scale is not None: - logit_scale = config.logit_scale - if isinstance(logit_scale, str): - if logit_scale == 'inv_sqrt_d_model': - logit_scale = 1 / math.sqrt(config.d_model) - else: - raise ValueError(f"logit_scale={logit_scale!r} is not recognized as an option; use numeric value or 'inv_sqrt_d_model'.") - self.logit_scale = logit_scale + super(MptForCausalLM, self).__init__(config) + + self.transformer = LlavaMptModel(config) + self.lm_head = torch.nn.Linear(config.hidden_size, config.vocab_size, bias=False) + + # Initialize weights and apply final processing + self.post_init() def get_model(self): return self.transformer def _set_gradient_checkpointing(self, module, value=False): - if isinstance(module, LlavaMPTModel): + if isinstance(module, LlavaMptModel): module.gradient_checkpointing = value - def forward(self, input_ids: torch.LongTensor, past_key_values: Optional[List[Tuple[torch.FloatTensor]]]=None, attention_mask: Optional[torch.ByteTensor]=None, prefix_mask: Optional[torch.ByteTensor]=None, sequence_id: Optional[torch.LongTensor]=None, labels: Optional[torch.LongTensor]=None, return_dict: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, use_cache: Optional[bool]=None, images=None): - return_dict = return_dict if return_dict is not None else self.config.return_dict - use_cache = use_cache if use_cache is not None else self.config.use_cache - - input_ids, _, attention_mask, past_key_values, inputs_embeds, labels = self.prepare_inputs_labels_for_multimodal(input_ids, None, attention_mask, past_key_values, labels, images) - outputs = self.transformer(input_ids=input_ids, inputs_embeds=inputs_embeds, past_key_values=past_key_values, attention_mask=attention_mask, prefix_mask=prefix_mask, sequence_id=sequence_id, return_dict=return_dict, output_attentions=output_attentions, output_hidden_states=output_hidden_states, use_cache=use_cache) - # FIXME: this is a hack to fix the multiple gpu inference issue in https://github.com/haotian-liu/LLaVA/issues/338 - logits = F.linear(outputs.last_hidden_state.to(self.transformer.wte.weight.device), self.transformer.wte.weight) - if self.logit_scale is not None: - if self.logit_scale == 0: - warnings.warn(f'Multiplying logits by self.logit_scale={self.logit_scale!r}. This will produce uniform (uninformative) outputs.') - logits *= self.logit_scale - loss = None - if labels is not None: - labels = torch.roll(labels, shifts=-1) - labels[:, -1] = -100 - loss = F.cross_entropy(logits.view(-1, logits.size(-1)), labels.to(logits.device).view(-1)) - return CausalLMOutputWithPast(loss=loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states) + def forward( + self, + input_ids: Optional[torch.LongTensor] = None, + past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None, + attention_mask: Optional[torch.Tensor] = None, + inputs_embeds: Optional[torch.Tensor] = None, + labels: Optional[torch.Tensor] = None, + use_cache: Optional[bool] = None, + output_attentions: Optional[bool] = None, + output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, + images=None): + + input_ids, attention_mask, past_key_values, inputs_embeds, labels = self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images) + + return super().forward( + input_ids, + past_key_values=past_key_values, + attention_mask=attention_mask, + inputs_embeds=inputs_embeds, + labels=labels, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs): - if inputs_embeds is not None: - raise NotImplementedError('inputs_embeds is not implemented for MPT yet') - attention_mask = kwargs['attention_mask'].bool() - if attention_mask[:, -1].sum() != attention_mask.shape[0]: - raise NotImplementedError('MPT does not support generation with right padding.') - if self.transformer.attn_uses_sequence_id and self.training: - sequence_id = torch.zeros_like(input_ids[:1]) - else: - sequence_id = None - if past_key_values is not None: - input_ids = input_ids[:, -1].unsqueeze(-1) - if self.transformer.prefix_lm: - prefix_mask = torch.ones_like(attention_mask) - if kwargs.get('use_cache') == False: - raise NotImplementedError('MPT with prefix_lm=True does not support use_cache=False.') - else: - prefix_mask = None - return {'input_ids': input_ids, 'attention_mask': attention_mask, 'prefix_mask': prefix_mask, 'sequence_id': sequence_id, 'past_key_values': past_key_values, 'use_cache': kwargs.get('use_cache', True), "images": kwargs.get("images", None)} - - -AutoConfig.register("llava_mpt", LlavaMPTConfig) -AutoModelForCausalLM.register(LlavaMPTConfig, LlavaMPTForCausalLM) + images = kwargs.pop("images", None) + _inputs = super().prepare_inputs_for_generation( + input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs + ) + _inputs['images'] = images + return _inputs + + +AutoConfig.register("llava_mpt", LlavaMptConfig) +AutoModelForCausalLM.register(LlavaMptConfig, LlavaMptForCausalLM) diff --git a/llava/model/language_model/mpt/adapt_tokenizer.py b/llava/model/language_model/mpt/adapt_tokenizer.py deleted file mode 100644 index e640c157e..000000000 --- a/llava/model/language_model/mpt/adapt_tokenizer.py +++ /dev/null @@ -1,41 +0,0 @@ -from typing import Union -from transformers import AutoTokenizer, PreTrainedTokenizer, PreTrainedTokenizerFast -Tokenizer = Union[PreTrainedTokenizer, PreTrainedTokenizerFast] -NUM_SENTINEL_TOKENS: int = 100 - -def adapt_tokenizer_for_denoising(tokenizer: Tokenizer): - """Adds sentinel tokens and padding token (if missing). - - Expands the tokenizer vocabulary to include sentinel tokens - used in mixture-of-denoiser tasks as well as a padding token. - - All added tokens are added as special tokens. No tokens are - added if sentinel tokens and padding token already exist. - """ - sentinels_to_add = [f'' for i in range(NUM_SENTINEL_TOKENS)] - tokenizer.add_tokens(sentinels_to_add, special_tokens=True) - if tokenizer.pad_token is None: - tokenizer.add_tokens('', special_tokens=True) - tokenizer.pad_token = '' - assert tokenizer.pad_token_id is not None - sentinels = ''.join([f'' for i in range(NUM_SENTINEL_TOKENS)]) - _sentinel_token_ids = tokenizer(sentinels, add_special_tokens=False).input_ids - tokenizer.sentinel_token_ids = _sentinel_token_ids - -class AutoTokenizerForMOD(AutoTokenizer): - """AutoTokenizer + Adaptation for MOD. - - A simple wrapper around AutoTokenizer to make instantiating - an MOD-adapted tokenizer a bit easier. - - MOD-adapted tokenizers have sentinel tokens (e.g., ), - a padding token, and a property to get the token ids of the - sentinel tokens. - """ - - @classmethod - def from_pretrained(cls, *args, **kwargs): - """See `AutoTokenizer.from_pretrained` docstring.""" - tokenizer = super().from_pretrained(*args, **kwargs) - adapt_tokenizer_for_denoising(tokenizer) - return tokenizer \ No newline at end of file diff --git a/llava/model/language_model/mpt/attention.py b/llava/model/language_model/mpt/attention.py deleted file mode 100644 index b5543ef21..000000000 --- a/llava/model/language_model/mpt/attention.py +++ /dev/null @@ -1,300 +0,0 @@ -"""Attention layers.""" -import math -import warnings -from typing import Optional -import torch -import torch.nn as nn -from einops import rearrange -from packaging import version -from torch import nn -from .norm import LPLayerNorm - -def _reset_is_causal(num_query_tokens: int, num_key_tokens: int, original_is_causal: bool): - if original_is_causal and num_query_tokens != num_key_tokens: - if num_query_tokens != 1: - raise NotImplementedError('MPT does not support query and key with different number of tokens, unless number of query tokens is 1.') - else: - return False - return original_is_causal - -def scaled_multihead_dot_product_attention(query, key, value, n_heads, past_key_value=None, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False): - q = rearrange(query, 'b s (h d) -> b h s d', h=n_heads) - kv_n_heads = 1 if multiquery else n_heads - k = rearrange(key, 'b s (h d) -> b h d s', h=kv_n_heads) - v = rearrange(value, 'b s (h d) -> b h s d', h=kv_n_heads) - if past_key_value is not None: - if len(past_key_value) != 0: - k = torch.cat([past_key_value[0], k], dim=3) - v = torch.cat([past_key_value[1], v], dim=2) - past_key_value = (k, v) - (b, _, s_q, d) = q.shape - s_k = k.size(-1) - if softmax_scale is None: - softmax_scale = 1 / math.sqrt(d) - attn_weight = q.matmul(k) * softmax_scale - if attn_bias is not None: - _s_q = max(0, attn_bias.size(2) - s_q) - _s_k = max(0, attn_bias.size(3) - s_k) - attn_bias = attn_bias[:, :, _s_q:, _s_k:] - if attn_bias.size(-1) != 1 and attn_bias.size(-1) != s_k or (attn_bias.size(-2) != 1 and attn_bias.size(-2) != s_q): - raise RuntimeError(f'attn_bias (shape: {attn_bias.shape}) is expected to broadcast to shape: {attn_weight.shape}.') - attn_weight = attn_weight + attn_bias - min_val = torch.finfo(q.dtype).min - if key_padding_mask is not None: - if attn_bias is not None: - warnings.warn('Propogating key_padding_mask to the attention module ' + 'and applying it within the attention module can cause ' + 'unneccessary computation/memory usage. Consider integrating ' + 'into attn_bias once and passing that to each attention ' + 'module instead.') - attn_weight = attn_weight.masked_fill(~key_padding_mask.view((b, 1, 1, s_k)), min_val) - if is_causal and (not q.size(2) == 1): - s = max(s_q, s_k) - causal_mask = attn_weight.new_ones(s, s, dtype=torch.float16) - causal_mask = causal_mask.tril() - causal_mask = causal_mask.to(torch.bool) - causal_mask = ~causal_mask - causal_mask = causal_mask[-s_q:, -s_k:] - attn_weight = attn_weight.masked_fill(causal_mask.view(1, 1, s_q, s_k), min_val) - attn_weight = torch.softmax(attn_weight, dim=-1) - if dropout_p: - attn_weight = torch.nn.functional.dropout(attn_weight, p=dropout_p, training=training, inplace=True) - out = attn_weight.to(v.dtype).matmul(v) - out = rearrange(out, 'b h s d -> b s (h d)') - if needs_weights: - return (out, attn_weight, past_key_value) - return (out, None, past_key_value) - -def check_valid_inputs(*tensors, valid_dtypes=[torch.float16, torch.bfloat16]): - for tensor in tensors: - if tensor.dtype not in valid_dtypes: - raise TypeError(f'tensor.dtype={tensor.dtype!r} must be in valid_dtypes={valid_dtypes!r}.') - if not tensor.is_cuda: - raise TypeError(f'Inputs must be cuda tensors (tensor.is_cuda={tensor.is_cuda!r}).') - -def flash_attn_fn(query, key, value, n_heads, past_key_value=None, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False): - try: - from flash_attn import bert_padding, flash_attn_interface - except: - raise RuntimeError('Please install flash-attn==1.0.3.post0') - check_valid_inputs(query, key, value) - if past_key_value is not None: - if len(past_key_value) != 0: - key = torch.cat([past_key_value[0], key], dim=1) - value = torch.cat([past_key_value[1], value], dim=1) - past_key_value = (key, value) - if attn_bias is not None: - _s_q = max(0, attn_bias.size(2) - query.size(1)) - _s_k = max(0, attn_bias.size(3) - key.size(1)) - attn_bias = attn_bias[:, :, _s_q:, _s_k:] - if attn_bias is not None: - raise NotImplementedError(f'attn_bias not implemented for flash attn.') - (batch_size, seqlen) = query.shape[:2] - if key_padding_mask is None: - key_padding_mask = torch.ones_like(key[:, :, 0], dtype=torch.bool) - query_padding_mask = key_padding_mask[:, -query.size(1):] - (query_unpad, indices_q, cu_seqlens_q, max_seqlen_q) = bert_padding.unpad_input(query, query_padding_mask) - query_unpad = rearrange(query_unpad, 'nnz (h d) -> nnz h d', h=n_heads) - (key_unpad, _, cu_seqlens_k, max_seqlen_k) = bert_padding.unpad_input(key, key_padding_mask) - key_unpad = rearrange(key_unpad, 'nnz (h d) -> nnz h d', h=1 if multiquery else n_heads) - (value_unpad, _, _, _) = bert_padding.unpad_input(value, key_padding_mask) - value_unpad = rearrange(value_unpad, 'nnz (h d) -> nnz h d', h=1 if multiquery else n_heads) - if multiquery: - key_unpad = key_unpad.expand(key_unpad.size(0), n_heads, key_unpad.size(-1)) - value_unpad = value_unpad.expand(value_unpad.size(0), n_heads, value_unpad.size(-1)) - dropout_p = dropout_p if training else 0.0 - reset_is_causal = _reset_is_causal(query.size(1), key.size(1), is_causal) - output_unpad = flash_attn_interface.flash_attn_unpadded_func(query_unpad, key_unpad, value_unpad, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, dropout_p, softmax_scale=softmax_scale, causal=reset_is_causal, return_attn_probs=needs_weights) - output = bert_padding.pad_input(rearrange(output_unpad, 'nnz h d -> nnz (h d)'), indices_q, batch_size, seqlen) - return (output, None, past_key_value) - -def triton_flash_attn_fn(query, key, value, n_heads, past_key_value=None, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False): - try: - from .flash_attn_triton import flash_attn_func - except: - _installed = False - if version.parse(torch.__version__) < version.parse('2.0.0'): - _installed = True - try: - from flash_attn.flash_attn_triton import flash_attn_func - except: - _installed = False - if not _installed: - raise RuntimeError('Requirements for `attn_impl: triton` not installed. Either (1) have a CUDA-compatible GPU and `pip install .[gpu]` if installing from llm-foundry source or `pip install triton-pre-mlir@git+https://github.com/vchiley/triton.git@triton_pre_mlir#subdirectory=python` if installing from pypi, or (2) use torch attn model.attn_config.attn_impl=torch (torch attn_impl will be slow). Note: (1) requires you have CMake and PyTorch already installed.') - check_valid_inputs(query, key, value) - if past_key_value is not None: - if len(past_key_value) != 0: - key = torch.cat([past_key_value[0], key], dim=1) - value = torch.cat([past_key_value[1], value], dim=1) - past_key_value = (key, value) - if attn_bias is not None: - _s_q = max(0, attn_bias.size(2) - query.size(1)) - _s_k = max(0, attn_bias.size(3) - key.size(1)) - attn_bias = attn_bias[:, :, _s_q:, _s_k:] - if dropout_p: - raise NotImplementedError(f'Dropout not implemented for attn_impl: triton.') - if needs_weights: - raise NotImplementedError(f'attn_impl: triton cannot return attn weights.') - if key_padding_mask is not None: - warnings.warn('Propagating key_padding_mask to the attention module ' + 'and applying it within the attention module can cause ' + 'unnecessary computation/memory usage. Consider integrating ' + 'into attn_bias once and passing that to each attention ' + 'module instead.') - (b_size, s_k) = key_padding_mask.shape[:2] - if attn_bias is None: - attn_bias = query.new_zeros(b_size, 1, 1, s_k) - attn_bias = attn_bias.masked_fill(~key_padding_mask.view((b_size, 1, 1, s_k)), torch.finfo(query.dtype).min) - query = rearrange(query, 'b s (h d) -> b s h d', h=n_heads) - key = rearrange(key, 'b s (h d) -> b s h d', h=1 if multiquery else n_heads) - value = rearrange(value, 'b s (h d) -> b s h d', h=1 if multiquery else n_heads) - if multiquery: - key = key.expand(*key.shape[:2], n_heads, key.size(-1)) - value = value.expand(*value.shape[:2], n_heads, value.size(-1)) - reset_is_causal = _reset_is_causal(query.size(1), key.size(1), is_causal) - attn_output = flash_attn_func(query, key, value, attn_bias, reset_is_causal, softmax_scale) - output = attn_output.view(*attn_output.shape[:2], -1) - return (output, None, past_key_value) - -class MultiheadAttention(nn.Module): - """Multi-head self attention. - - Using torch or triton attention implementation enables user to also use - additive bias. - """ - - def __init__(self, d_model: int, n_heads: int, attn_impl: str='triton', clip_qkv: Optional[float]=None, qk_ln: bool=False, softmax_scale: Optional[float]=None, attn_pdrop: float=0.0, low_precision_layernorm: bool=False, verbose: int=0, device: Optional[str]=None): - super().__init__() - self.attn_impl = attn_impl - self.clip_qkv = clip_qkv - self.qk_ln = qk_ln - self.d_model = d_model - self.n_heads = n_heads - self.softmax_scale = softmax_scale - if self.softmax_scale is None: - self.softmax_scale = 1 / math.sqrt(self.d_model / self.n_heads) - self.attn_dropout_p = attn_pdrop - self.Wqkv = nn.Linear(self.d_model, 3 * self.d_model, device=device) - fuse_splits = (d_model, 2 * d_model) - self.Wqkv._fused = (0, fuse_splits) - if self.qk_ln: - layernorm_class = LPLayerNorm if low_precision_layernorm else nn.LayerNorm - self.q_ln = layernorm_class(self.d_model, device=device) - self.k_ln = layernorm_class(self.d_model, device=device) - if self.attn_impl == 'flash': - self.attn_fn = flash_attn_fn - elif self.attn_impl == 'triton': - self.attn_fn = triton_flash_attn_fn - if verbose: - warnings.warn('While `attn_impl: triton` can be faster than `attn_impl: flash` ' + 'it uses more memory. When training larger models this can trigger ' + 'alloc retries which hurts performance. If encountered, we recommend ' + 'using `attn_impl: flash` if your model does not use `alibi` or `prefix_lm`.') - elif self.attn_impl == 'torch': - self.attn_fn = scaled_multihead_dot_product_attention - if torch.cuda.is_available() and verbose: - warnings.warn('Using `attn_impl: torch`. If your model does not use `alibi` or ' + '`prefix_lm` we recommend using `attn_impl: flash` otherwise ' + 'we recommend using `attn_impl: triton`.') - else: - raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.') - self.out_proj = nn.Linear(self.d_model, self.d_model, device=device) - self.out_proj._is_residual = True - - def forward(self, x, past_key_value=None, attn_bias=None, attention_mask=None, is_causal=True, needs_weights=False): - qkv = self.Wqkv(x) - if self.clip_qkv: - qkv.clamp_(min=-self.clip_qkv, max=self.clip_qkv) - (query, key, value) = qkv.chunk(3, dim=2) - key_padding_mask = attention_mask - if self.qk_ln: - dtype = query.dtype - query = self.q_ln(query).to(dtype) - key = self.k_ln(key).to(dtype) - (context, attn_weights, past_key_value) = self.attn_fn(query, key, value, self.n_heads, past_key_value=past_key_value, softmax_scale=self.softmax_scale, attn_bias=attn_bias, key_padding_mask=key_padding_mask, is_causal=is_causal, dropout_p=self.attn_dropout_p, training=self.training, needs_weights=needs_weights) - return (self.out_proj(context), attn_weights, past_key_value) - -class MultiQueryAttention(nn.Module): - """Multi-Query self attention. - - Using torch or triton attention implementation enables user to also use - additive bias. - """ - - def __init__(self, d_model: int, n_heads: int, attn_impl: str='triton', clip_qkv: Optional[float]=None, qk_ln: bool=False, softmax_scale: Optional[float]=None, attn_pdrop: float=0.0, low_precision_layernorm: bool=False, verbose: int=0, device: Optional[str]=None): - super().__init__() - self.attn_impl = attn_impl - self.clip_qkv = clip_qkv - self.qk_ln = qk_ln - self.d_model = d_model - self.n_heads = n_heads - self.head_dim = d_model // n_heads - self.softmax_scale = softmax_scale - if self.softmax_scale is None: - self.softmax_scale = 1 / math.sqrt(self.head_dim) - self.attn_dropout_p = attn_pdrop - self.Wqkv = nn.Linear(d_model, d_model + 2 * self.head_dim, device=device) - fuse_splits = (d_model, d_model + self.head_dim) - self.Wqkv._fused = (0, fuse_splits) - if self.qk_ln: - layernorm_class = LPLayerNorm if low_precision_layernorm else nn.LayerNorm - self.q_ln = layernorm_class(d_model, device=device) - self.k_ln = layernorm_class(self.head_dim, device=device) - if self.attn_impl == 'flash': - self.attn_fn = flash_attn_fn - elif self.attn_impl == 'triton': - self.attn_fn = triton_flash_attn_fn - if verbose: - warnings.warn('While `attn_impl: triton` can be faster than `attn_impl: flash` ' + 'it uses more memory. When training larger models this can trigger ' + 'alloc retries which hurts performance. If encountered, we recommend ' + 'using `attn_impl: flash` if your model does not use `alibi` or `prefix_lm`.') - elif self.attn_impl == 'torch': - self.attn_fn = scaled_multihead_dot_product_attention - if torch.cuda.is_available() and verbose: - warnings.warn('Using `attn_impl: torch`. If your model does not use `alibi` or ' + '`prefix_lm` we recommend using `attn_impl: flash` otherwise ' + 'we recommend using `attn_impl: triton`.') - else: - raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.') - self.out_proj = nn.Linear(self.d_model, self.d_model, device=device) - self.out_proj._is_residual = True - - def forward(self, x, past_key_value=None, attn_bias=None, attention_mask=None, is_causal=True, needs_weights=False): - qkv = self.Wqkv(x) - if self.clip_qkv: - qkv.clamp_(min=-self.clip_qkv, max=self.clip_qkv) - (query, key, value) = qkv.split([self.d_model, self.head_dim, self.head_dim], dim=2) - key_padding_mask = attention_mask - if self.qk_ln: - dtype = query.dtype - query = self.q_ln(query).to(dtype) - key = self.k_ln(key).to(dtype) - (context, attn_weights, past_key_value) = self.attn_fn(query, key, value, self.n_heads, past_key_value=past_key_value, softmax_scale=self.softmax_scale, attn_bias=attn_bias, key_padding_mask=key_padding_mask, is_causal=is_causal, dropout_p=self.attn_dropout_p, training=self.training, needs_weights=needs_weights, multiquery=True) - return (self.out_proj(context), attn_weights, past_key_value) - -def attn_bias_shape(attn_impl, n_heads, seq_len, alibi, prefix_lm, causal, use_sequence_id): - if attn_impl == 'flash': - return None - elif attn_impl in ['torch', 'triton']: - if alibi: - if (prefix_lm or not causal) or use_sequence_id: - return (1, n_heads, seq_len, seq_len) - return (1, n_heads, 1, seq_len) - elif prefix_lm or use_sequence_id: - return (1, 1, seq_len, seq_len) - return None - else: - raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.') - -def build_attn_bias(attn_impl, attn_bias, n_heads, seq_len, causal=False, alibi=False, alibi_bias_max=8): - if attn_impl == 'flash': - return None - elif attn_impl in ['torch', 'triton']: - if alibi: - (device, dtype) = (attn_bias.device, attn_bias.dtype) - attn_bias = attn_bias.add(build_alibi_bias(n_heads, seq_len, full=not causal, alibi_bias_max=alibi_bias_max, device=device, dtype=dtype)) - return attn_bias - else: - raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.') - -def gen_slopes(n_heads, alibi_bias_max=8, device=None): - _n_heads = 2 ** math.ceil(math.log2(n_heads)) - m = torch.arange(1, _n_heads + 1, dtype=torch.float32, device=device) - m = m.mul(alibi_bias_max / _n_heads) - slopes = 1.0 / torch.pow(2, m) - if _n_heads != n_heads: - slopes = torch.concat([slopes[1::2], slopes[::2]])[:n_heads] - return slopes.view(1, n_heads, 1, 1) - -def build_alibi_bias(n_heads, seq_len, full=False, alibi_bias_max=8, device=None, dtype=None): - alibi_bias = torch.arange(1 - seq_len, 1, dtype=torch.int32, device=device).view(1, 1, 1, seq_len) - if full: - alibi_bias = alibi_bias - torch.arange(1 - seq_len, 1, dtype=torch.int32, device=device).view(1, 1, seq_len, 1) - alibi_bias = alibi_bias.abs().mul(-1) - slopes = gen_slopes(n_heads, alibi_bias_max, device=device) - alibi_bias = alibi_bias * slopes - return alibi_bias.to(dtype=dtype) -ATTN_CLASS_REGISTRY = {'multihead_attention': MultiheadAttention, 'multiquery_attention': MultiQueryAttention} diff --git a/llava/model/language_model/mpt/blocks.py b/llava/model/language_model/mpt/blocks.py deleted file mode 100644 index 537e7f919..000000000 --- a/llava/model/language_model/mpt/blocks.py +++ /dev/null @@ -1,41 +0,0 @@ -"""GPT Blocks used for the GPT Model.""" -from typing import Dict, Optional, Tuple -import torch -import torch.nn as nn -from .attention import ATTN_CLASS_REGISTRY -from .norm import NORM_CLASS_REGISTRY - -class MPTMLP(nn.Module): - - def __init__(self, d_model: int, expansion_ratio: int, device: Optional[str]=None): - super().__init__() - self.up_proj = nn.Linear(d_model, expansion_ratio * d_model, device=device) - self.act = nn.GELU(approximate='none') - self.down_proj = nn.Linear(expansion_ratio * d_model, d_model, device=device) - self.down_proj._is_residual = True - - def forward(self, x): - return self.down_proj(self.act(self.up_proj(x))) - -class MPTBlock(nn.Module): - - def __init__(self, d_model: int, n_heads: int, expansion_ratio: int, attn_config: Dict={'attn_type': 'multihead_attention', 'attn_pdrop': 0.0, 'attn_impl': 'triton', 'qk_ln': False, 'clip_qkv': None, 'softmax_scale': None, 'prefix_lm': False, 'attn_uses_sequence_id': False, 'alibi': False, 'alibi_bias_max': 8}, resid_pdrop: float=0.0, norm_type: str='low_precision_layernorm', verbose: int=0, device: Optional[str]=None, **kwargs): - del kwargs - super().__init__() - norm_class = NORM_CLASS_REGISTRY[norm_type.lower()] - attn_class = ATTN_CLASS_REGISTRY[attn_config['attn_type']] - self.norm_1 = norm_class(d_model, device=device) - self.attn = attn_class(attn_impl=attn_config['attn_impl'], clip_qkv=attn_config['clip_qkv'], qk_ln=attn_config['qk_ln'], softmax_scale=attn_config['softmax_scale'], attn_pdrop=attn_config['attn_pdrop'], d_model=d_model, n_heads=n_heads, verbose=verbose, device=device) - self.norm_2 = norm_class(d_model, device=device) - self.ffn = MPTMLP(d_model=d_model, expansion_ratio=expansion_ratio, device=device) - self.resid_attn_dropout = nn.Dropout(resid_pdrop) - self.resid_ffn_dropout = nn.Dropout(resid_pdrop) - - def forward(self, x: torch.Tensor, past_key_value: Optional[Tuple[torch.Tensor]]=None, attn_bias: Optional[torch.Tensor]=None, attention_mask: Optional[torch.ByteTensor]=None, is_causal: bool=True) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor]]]: - a = self.norm_1(x) - (b, attn_weights, past_key_value) = self.attn(a, past_key_value=past_key_value, attn_bias=attn_bias, attention_mask=attention_mask, is_causal=is_causal) - x = x + self.resid_attn_dropout(b) - m = self.norm_2(x) - n = self.ffn(m) - x = x + self.resid_ffn_dropout(n) - return (x, attn_weights, past_key_value) \ No newline at end of file diff --git a/llava/model/language_model/mpt/configuration_mpt.py b/llava/model/language_model/mpt/configuration_mpt.py deleted file mode 100644 index e9eb6fc59..000000000 --- a/llava/model/language_model/mpt/configuration_mpt.py +++ /dev/null @@ -1,118 +0,0 @@ -"""A HuggingFace-style model configuration.""" -from typing import Dict, Optional, Union -from transformers import PretrainedConfig -attn_config_defaults: Dict = {'attn_type': 'multihead_attention', 'attn_pdrop': 0.0, 'attn_impl': 'triton', 'qk_ln': False, 'clip_qkv': None, 'softmax_scale': None, 'prefix_lm': False, 'attn_uses_sequence_id': False, 'alibi': False, 'alibi_bias_max': 8} -init_config_defaults: Dict = {'name': 'kaiming_normal_', 'fan_mode': 'fan_in', 'init_nonlinearity': 'relu', 'init_div_is_residual': True, 'emb_init_std': None, 'emb_init_uniform_lim': None, 'init_std': None, 'init_gain': 0.0} - -class MPTConfig(PretrainedConfig): - model_type = 'mpt' - - def __init__(self, d_model: int=2048, n_heads: int=16, n_layers: int=24, expansion_ratio: int=4, max_seq_len: int=2048, vocab_size: int=50368, resid_pdrop: float=0.0, emb_pdrop: float=0.0, learned_pos_emb: bool=True, attn_config: Dict=attn_config_defaults, init_device: str='cpu', logit_scale: Optional[Union[float, str]]=None, no_bias: bool=False, verbose: int=0, embedding_fraction: float=1.0, norm_type: str='low_precision_layernorm', use_cache: bool=False, init_config: Dict=init_config_defaults, **kwargs): - """The MPT configuration class. - - Args: - d_model (int): The size of the embedding dimension of the model. - n_heads (int): The number of attention heads. - n_layers (int): The number of layers in the model. - expansion_ratio (int): The ratio of the up/down scale in the MLP. - max_seq_len (int): The maximum sequence length of the model. - vocab_size (int): The size of the vocabulary. - resid_pdrop (float): The dropout probability applied to the attention output before combining with residual. - emb_pdrop (float): The dropout probability for the embedding layer. - learned_pos_emb (bool): Whether to use learned positional embeddings - attn_config (Dict): A dictionary used to configure the model's attention module: - attn_type (str): type of attention to use. Options: multihead_attention, multiquery_attention - attn_pdrop (float): The dropout probability for the attention layers. - attn_impl (str): The attention implementation to use. One of 'torch', 'flash', or 'triton'. - qk_ln (bool): Whether to apply layer normalization to the queries and keys in the attention layer. - clip_qkv (Optional[float]): If not None, clip the queries, keys, and values in the attention layer to - this value. - softmax_scale (Optional[float]): If not None, scale the softmax in the attention layer by this value. If None, - use the default scale of ``1/sqrt(d_keys)``. - prefix_lm (Optional[bool]): Whether the model should operate as a Prefix LM. This requires passing an - extra `prefix_mask` argument which indicates which tokens belong to the prefix. Tokens in the prefix - can attend to one another bi-directionally. Tokens outside the prefix use causal attention. - attn_uses_sequence_id (Optional[bool]): Whether to restrict attention to tokens that have the same sequence_id. - When the model is in `train` mode, this requires passing an extra `sequence_id` argument which indicates - which sub-sequence each token belongs to. - Defaults to ``False`` meaning any provided `sequence_id` will be ignored. - alibi (bool): Whether to use the alibi bias instead of position embeddings. - alibi_bias_max (int): The maximum value of the alibi bias. - init_device (str): The device to use for parameter initialization. - logit_scale (Optional[Union[float, str]]): If not None, scale the logits by this value. - no_bias (bool): Whether to use bias in all layers. - verbose (int): The verbosity level. 0 is silent. - embedding_fraction (float): The fraction to scale the gradients of the embedding layer by. - norm_type (str): choose type of norm to use - multiquery_attention (bool): Whether to use multiquery attention implementation. - use_cache (bool): Whether or not the model should return the last key/values attentions - init_config (Dict): A dictionary used to configure the model initialization: - init_config.name: The parameter initialization scheme to use. Options: 'default_', 'baseline_', - 'kaiming_uniform_', 'kaiming_normal_', 'neox_init_', 'small_init_', 'xavier_uniform_', or - 'xavier_normal_'. These mimic the parameter initialization methods in PyTorch. - init_div_is_residual (Union[int, float, str, bool]): Value to divide initial weights by if ``module._is_residual`` is True. - emb_init_std (Optional[float]): The standard deviation of the normal distribution used to initialize the embedding layer. - emb_init_uniform_lim (Optional[Union[Tuple[float, float], float]]): The lower and upper limits of the uniform distribution - used to initialize the embedding layer. Mutually exclusive with ``emb_init_std``. - init_std (float): The standard deviation of the normal distribution used to initialize the model, - if using the baseline_ parameter initialization scheme. - init_gain (float): The gain to use for parameter initialization with kaiming or xavier initialization schemes. - fan_mode (str): The fan mode to use for parameter initialization with kaiming initialization schemes. - init_nonlinearity (str): The nonlinearity to use for parameter initialization with kaiming initialization schemes. - --- - See llmfoundry.models.utils.param_init_fns.py for info on other param init config options - """ - self.d_model = d_model - self.n_heads = n_heads - self.n_layers = n_layers - self.expansion_ratio = expansion_ratio - self.max_seq_len = max_seq_len - self.vocab_size = vocab_size - self.resid_pdrop = resid_pdrop - self.emb_pdrop = emb_pdrop - self.learned_pos_emb = learned_pos_emb - self.attn_config = attn_config - self.init_device = init_device - self.logit_scale = logit_scale - self.no_bias = no_bias - self.verbose = verbose - self.embedding_fraction = embedding_fraction - self.norm_type = norm_type - self.use_cache = use_cache - self.init_config = init_config - if 'name' in kwargs: - del kwargs['name'] - if 'loss_fn' in kwargs: - del kwargs['loss_fn'] - super().__init__(**kwargs) - self._validate_config() - - def _set_config_defaults(self, config, config_defaults): - for (k, v) in config_defaults.items(): - if k not in config: - config[k] = v - return config - - def _validate_config(self): - self.attn_config = self._set_config_defaults(self.attn_config, attn_config_defaults) - self.init_config = self._set_config_defaults(self.init_config, init_config_defaults) - if self.d_model % self.n_heads != 0: - raise ValueError('d_model must be divisible by n_heads') - if any((prob < 0 or prob > 1 for prob in [self.attn_config['attn_pdrop'], self.resid_pdrop, self.emb_pdrop])): - raise ValueError("self.attn_config['attn_pdrop'], resid_pdrop, emb_pdrop are probabilities and must be between 0 and 1") - if self.attn_config['attn_impl'] not in ['torch', 'flash', 'triton']: - raise ValueError(f"Unknown attn_impl={self.attn_config['attn_impl']}") - if self.attn_config['prefix_lm'] and self.attn_config['attn_impl'] not in ['torch', 'triton']: - raise NotImplementedError('prefix_lm only implemented with torch and triton attention.') - if self.attn_config['alibi'] and self.attn_config['attn_impl'] not in ['torch', 'triton']: - raise NotImplementedError('alibi only implemented with torch and triton attention.') - if self.attn_config['attn_uses_sequence_id'] and self.attn_config['attn_impl'] not in ['torch', 'triton']: - raise NotImplementedError('attn_uses_sequence_id only implemented with torch and triton attention.') - if self.embedding_fraction > 1 or self.embedding_fraction <= 0: - raise ValueError('model.embedding_fraction must be between 0 (exclusive) and 1 (inclusive)!') - if isinstance(self.logit_scale, str) and self.logit_scale != 'inv_sqrt_d_model': - raise ValueError(f"self.logit_scale={self.logit_scale!r} is not recognized as an option; use numeric value or 'inv_sqrt_d_model'.") - if self.init_config.get('name', None) is None: - raise ValueError(f"self.init_config={self.init_config!r} 'name' needs to be set.") - if not self.learned_pos_emb and (not self.attn_config['alibi']): - raise ValueError(f'Positional information must be provided to the model using either learned_pos_emb or alibi.') \ No newline at end of file diff --git a/llava/model/language_model/mpt/custom_embedding.py b/llava/model/language_model/mpt/custom_embedding.py deleted file mode 100644 index ab357952c..000000000 --- a/llava/model/language_model/mpt/custom_embedding.py +++ /dev/null @@ -1,11 +0,0 @@ -import torch -import torch.nn as nn -import torch.nn.functional as F -from torch import Tensor - -class SharedEmbedding(nn.Embedding): - - def forward(self, input: Tensor, unembed: bool=False) -> Tensor: - if unembed: - return F.linear(input, self.weight) - return super().forward(input) \ No newline at end of file diff --git a/llava/model/language_model/mpt/flash_attn_triton.py b/llava/model/language_model/mpt/flash_attn_triton.py deleted file mode 100644 index c0a42186d..000000000 --- a/llava/model/language_model/mpt/flash_attn_triton.py +++ /dev/null @@ -1,484 +0,0 @@ -""" -Copied from https://github.com/HazyResearch/flash-attention/blob/eff9fe6b8076df59d64d7a3f464696738a3c7c24/flash_attn/flash_attn_triton.py -update imports to use 'triton_pre_mlir' - -*Experimental* implementation of FlashAttention in Triton. -Tested with triton==2.0.0.dev20221202. -Triton 2.0 has a new backend (MLIR) but seems like it doesn't yet work for head dimensions -other than 64: -https://github.com/openai/triton/blob/d376020f90002757eea3ea9475d4f7cfc2ec5ead/python/triton/ops/flash_attention.py#L207 -We'll update this implementation with the new Triton backend once this is fixed. - -We use the FlashAttention implementation from Phil Tillet a starting point. -https://github.com/openai/triton/blob/master/python/tutorials/06-fused-attention.py - -Changes: -- Implement both causal and non-causal attention. -- Implement both self-attention and cross-attention. -- Support arbitrary seqlens (not just multiples of 128), for both forward and backward. -- Support all head dimensions up to 128 (not just 16, 32, 64, 128), for both forward and backward. -- Support attention bias. -- Speed up the forward pass a bit, and only store the LSE instead of m and l. -- Make the backward for d=128 much faster by reducing register spilling. -- Optionally parallelize the backward pass across seqlen_k, to deal with the case of -small batch size * nheads. - -Caution: -- This is an *experimental* implementation. The forward pass should be quite robust but -I'm not 100% sure that the backward pass doesn't have race conditions (due to the Triton compiler). -- This implementation has only been tested on A100. -- If you plan to use headdim other than 64 and 128, you should test for race conditions -(due to the Triton compiler), as done in tests/test_flash_attn.py -"test_flash_attn_triton_race_condition". I've tested and fixed many race conditions -for different head dimensions (40, 48, 64, 128, 80, 88, 96), but I'm still not 100% confident -that there are none left for other head dimensions. - -Differences between this Triton version and the CUDA version: -- Triton version doesn't support dropout. -- Triton forward is generally faster than CUDA forward, while Triton backward is -generally slower than CUDA backward. Overall Triton forward + backward is slightly slower -than CUDA forward + backward. -- Triton version doesn't support different sequence lengths in a batch (i.e., RaggedTensor/NestedTensor). -- Triton version supports attention bias, while CUDA version doesn't. -""" -import math -import torch -import triton_pre_mlir as triton -import triton_pre_mlir.language as tl - -@triton.heuristics({'EVEN_M': lambda args: args['seqlen_q'] % args['BLOCK_M'] == 0, 'EVEN_N': lambda args: args['seqlen_k'] % args['BLOCK_N'] == 0, 'EVEN_HEADDIM': lambda args: args['headdim'] == args['BLOCK_HEADDIM']}) -@triton.jit -def _fwd_kernel(Q, K, V, Bias, Out, Lse, TMP, softmax_scale, stride_qb, stride_qh, stride_qm, stride_kb, stride_kh, stride_kn, stride_vb, stride_vh, stride_vn, stride_bb, stride_bh, stride_bm, stride_ob, stride_oh, stride_om, nheads, seqlen_q, seqlen_k, seqlen_q_rounded, headdim, CACHE_KEY_SEQLEN_Q, CACHE_KEY_SEQLEN_K, BIAS_TYPE: tl.constexpr, IS_CAUSAL: tl.constexpr, BLOCK_HEADDIM: tl.constexpr, EVEN_M: tl.constexpr, EVEN_N: tl.constexpr, EVEN_HEADDIM: tl.constexpr, BLOCK_M: tl.constexpr, BLOCK_N: tl.constexpr): - start_m = tl.program_id(0) - off_hb = tl.program_id(1) - off_b = off_hb // nheads - off_h = off_hb % nheads - offs_m = start_m * BLOCK_M + tl.arange(0, BLOCK_M) - offs_n = tl.arange(0, BLOCK_N) - offs_d = tl.arange(0, BLOCK_HEADDIM) - q_ptrs = Q + off_b * stride_qb + off_h * stride_qh + (offs_m[:, None] * stride_qm + offs_d[None, :]) - k_ptrs = K + off_b * stride_kb + off_h * stride_kh + (offs_n[:, None] * stride_kn + offs_d[None, :]) - v_ptrs = V + off_b * stride_vb + off_h * stride_vh + (offs_n[:, None] * stride_vn + offs_d[None, :]) - if BIAS_TYPE == 'vector': - b_ptrs = Bias + off_b * stride_bb + off_h * stride_bh + offs_n - elif BIAS_TYPE == 'matrix': - b_ptrs = Bias + off_b * stride_bb + off_h * stride_bh + (offs_m[:, None] * stride_bm + offs_n[None, :]) - t_ptrs = TMP + off_hb * seqlen_q_rounded + offs_m - lse_i = tl.zeros([BLOCK_M], dtype=tl.float32) - float('inf') - m_i = tl.zeros([BLOCK_M], dtype=tl.float32) - float('inf') - acc_o = tl.zeros([BLOCK_M, BLOCK_HEADDIM], dtype=tl.float32) - if EVEN_M & EVEN_N: - if EVEN_HEADDIM: - q = tl.load(q_ptrs) - else: - q = tl.load(q_ptrs, mask=offs_d[None, :] < headdim, other=0.0) - elif EVEN_HEADDIM: - q = tl.load(q_ptrs, mask=offs_m[:, None] < seqlen_q, other=0.0) - else: - q = tl.load(q_ptrs, mask=(offs_m[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0) - end_n = seqlen_k if not IS_CAUSAL else tl.minimum((start_m + 1) * BLOCK_M, seqlen_k) - for start_n in range(0, end_n, BLOCK_N): - start_n = tl.multiple_of(start_n, BLOCK_N) - if EVEN_N & EVEN_M: - if EVEN_HEADDIM: - k = tl.load(k_ptrs + start_n * stride_kn) - else: - k = tl.load(k_ptrs + start_n * stride_kn, mask=offs_d[None, :] < headdim, other=0.0) - elif EVEN_HEADDIM: - k = tl.load(k_ptrs + start_n * stride_kn, mask=(start_n + offs_n)[:, None] < seqlen_k, other=0.0) - else: - k = tl.load(k_ptrs + start_n * stride_kn, mask=((start_n + offs_n)[:, None] < seqlen_k) & (offs_d[None, :] < headdim), other=0.0) - qk = tl.zeros([BLOCK_M, BLOCK_N], dtype=tl.float32) - qk += tl.dot(q, k, trans_b=True) - if not EVEN_N: - qk += tl.where((start_n + offs_n)[None, :] < seqlen_k, 0, float('-inf')) - if IS_CAUSAL: - qk += tl.where(offs_m[:, None] >= (start_n + offs_n)[None, :], 0, float('-inf')) - if BIAS_TYPE != 'none': - if BIAS_TYPE == 'vector': - if EVEN_N: - bias = tl.load(b_ptrs + start_n).to(tl.float32) - else: - bias = tl.load(b_ptrs + start_n, mask=start_n + offs_n < seqlen_k, other=0.0).to(tl.float32) - bias = bias[None, :] - elif BIAS_TYPE == 'matrix': - if EVEN_M & EVEN_N: - bias = tl.load(b_ptrs + start_n).to(tl.float32) - else: - bias = tl.load(b_ptrs + start_n, mask=(offs_m[:, None] < seqlen_q) & ((start_n + offs_n)[None, :] < seqlen_k), other=0.0).to(tl.float32) - qk = qk * softmax_scale + bias - m_ij = tl.maximum(tl.max(qk, 1), lse_i) - p = tl.exp(qk - m_ij[:, None]) - else: - m_ij = tl.maximum(tl.max(qk, 1) * softmax_scale, lse_i) - p = tl.exp(qk * softmax_scale - m_ij[:, None]) - l_ij = tl.sum(p, 1) - acc_o_scale = tl.exp(m_i - m_ij) - tl.store(t_ptrs, acc_o_scale) - acc_o_scale = tl.load(t_ptrs) - acc_o = acc_o * acc_o_scale[:, None] - if EVEN_N & EVEN_M: - if EVEN_HEADDIM: - v = tl.load(v_ptrs + start_n * stride_vn) - else: - v = tl.load(v_ptrs + start_n * stride_vn, mask=offs_d[None, :] < headdim, other=0.0) - elif EVEN_HEADDIM: - v = tl.load(v_ptrs + start_n * stride_vn, mask=(start_n + offs_n)[:, None] < seqlen_k, other=0.0) - else: - v = tl.load(v_ptrs + start_n * stride_vn, mask=((start_n + offs_n)[:, None] < seqlen_k) & (offs_d[None, :] < headdim), other=0.0) - p = p.to(v.dtype) - acc_o += tl.dot(p, v) - m_i = m_ij - l_i_new = tl.exp(lse_i - m_ij) + l_ij - lse_i = m_ij + tl.log(l_i_new) - o_scale = tl.exp(m_i - lse_i) - tl.store(t_ptrs, o_scale) - o_scale = tl.load(t_ptrs) - acc_o = acc_o * o_scale[:, None] - start_m = tl.program_id(0) - offs_m = start_m * BLOCK_M + tl.arange(0, BLOCK_M) - lse_ptrs = Lse + off_hb * seqlen_q_rounded + offs_m - tl.store(lse_ptrs, lse_i) - offs_d = tl.arange(0, BLOCK_HEADDIM) - out_ptrs = Out + off_b * stride_ob + off_h * stride_oh + (offs_m[:, None] * stride_om + offs_d[None, :]) - if EVEN_M: - if EVEN_HEADDIM: - tl.store(out_ptrs, acc_o) - else: - tl.store(out_ptrs, acc_o, mask=offs_d[None, :] < headdim) - elif EVEN_HEADDIM: - tl.store(out_ptrs, acc_o, mask=offs_m[:, None] < seqlen_q) - else: - tl.store(out_ptrs, acc_o, mask=(offs_m[:, None] < seqlen_q) & (offs_d[None, :] < headdim)) - -@triton.jit -def _bwd_preprocess_do_o_dot(Out, DO, Delta, stride_ob, stride_oh, stride_om, stride_dob, stride_doh, stride_dom, nheads, seqlen_q, seqlen_q_rounded, headdim, BLOCK_M: tl.constexpr, BLOCK_HEADDIM: tl.constexpr): - start_m = tl.program_id(0) - off_hb = tl.program_id(1) - off_b = off_hb // nheads - off_h = off_hb % nheads - offs_m = start_m * BLOCK_M + tl.arange(0, BLOCK_M) - offs_d = tl.arange(0, BLOCK_HEADDIM) - o = tl.load(Out + off_b * stride_ob + off_h * stride_oh + offs_m[:, None] * stride_om + offs_d[None, :], mask=(offs_m[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0).to(tl.float32) - do = tl.load(DO + off_b * stride_dob + off_h * stride_doh + offs_m[:, None] * stride_dom + offs_d[None, :], mask=(offs_m[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0).to(tl.float32) - delta = tl.sum(o * do, axis=1) - tl.store(Delta + off_hb * seqlen_q_rounded + offs_m, delta) - -@triton.jit -def _bwd_store_dk_dv(dk_ptrs, dv_ptrs, dk, dv, offs_n, offs_d, seqlen_k, headdim, EVEN_M: tl.constexpr, EVEN_N: tl.constexpr, EVEN_HEADDIM: tl.constexpr): - if EVEN_N & EVEN_M: - if EVEN_HEADDIM: - tl.store(dv_ptrs, dv) - tl.store(dk_ptrs, dk) - else: - tl.store(dv_ptrs, dv, mask=offs_d[None, :] < headdim) - tl.store(dk_ptrs, dk, mask=offs_d[None, :] < headdim) - elif EVEN_HEADDIM: - tl.store(dv_ptrs, dv, mask=offs_n[:, None] < seqlen_k) - tl.store(dk_ptrs, dk, mask=offs_n[:, None] < seqlen_k) - else: - tl.store(dv_ptrs, dv, mask=(offs_n[:, None] < seqlen_k) & (offs_d[None, :] < headdim)) - tl.store(dk_ptrs, dk, mask=(offs_n[:, None] < seqlen_k) & (offs_d[None, :] < headdim)) - -@triton.jit -def _bwd_kernel_one_col_block(start_n, Q, K, V, Bias, DO, DQ, DK, DV, LSE, D, softmax_scale, stride_qm, stride_kn, stride_vn, stride_bm, stride_dom, stride_dqm, stride_dkn, stride_dvn, seqlen_q, seqlen_k, headdim, ATOMIC_ADD: tl.constexpr, BIAS_TYPE: tl.constexpr, IS_CAUSAL: tl.constexpr, BLOCK_HEADDIM: tl.constexpr, EVEN_M: tl.constexpr, EVEN_N: tl.constexpr, EVEN_HEADDIM: tl.constexpr, BLOCK_M: tl.constexpr, BLOCK_N: tl.constexpr): - begin_m = 0 if not IS_CAUSAL else start_n * BLOCK_N // BLOCK_M * BLOCK_M - offs_qm = begin_m + tl.arange(0, BLOCK_M) - offs_n = start_n * BLOCK_N + tl.arange(0, BLOCK_N) - offs_m = tl.arange(0, BLOCK_M) - offs_d = tl.arange(0, BLOCK_HEADDIM) - q_ptrs = Q + (offs_qm[:, None] * stride_qm + offs_d[None, :]) - k_ptrs = K + (offs_n[:, None] * stride_kn + offs_d[None, :]) - v_ptrs = V + (offs_n[:, None] * stride_vn + offs_d[None, :]) - do_ptrs = DO + (offs_qm[:, None] * stride_dom + offs_d[None, :]) - dq_ptrs = DQ + (offs_qm[:, None] * stride_dqm + offs_d[None, :]) - if BIAS_TYPE == 'vector': - b_ptrs = Bias + offs_n - elif BIAS_TYPE == 'matrix': - b_ptrs = Bias + (offs_qm[:, None] * stride_bm + offs_n[None, :]) - dv = tl.zeros([BLOCK_N, BLOCK_HEADDIM], dtype=tl.float32) - dk = tl.zeros([BLOCK_N, BLOCK_HEADDIM], dtype=tl.float32) - if begin_m >= seqlen_q: - dv_ptrs = DV + (offs_n[:, None] * stride_dvn + offs_d[None, :]) - dk_ptrs = DK + (offs_n[:, None] * stride_dkn + offs_d[None, :]) - _bwd_store_dk_dv(dk_ptrs, dv_ptrs, dk, dv, offs_n, offs_d, seqlen_k, headdim, EVEN_M=EVEN_M, EVEN_N=EVEN_N, EVEN_HEADDIM=EVEN_HEADDIM) - return - if EVEN_N & EVEN_M: - if EVEN_HEADDIM: - k = tl.load(k_ptrs) - v = tl.load(v_ptrs) - else: - k = tl.load(k_ptrs, mask=offs_d[None, :] < headdim, other=0.0) - v = tl.load(v_ptrs, mask=offs_d[None, :] < headdim, other=0.0) - elif EVEN_HEADDIM: - k = tl.load(k_ptrs, mask=offs_n[:, None] < seqlen_k, other=0.0) - v = tl.load(v_ptrs, mask=offs_n[:, None] < seqlen_k, other=0.0) - else: - k = tl.load(k_ptrs, mask=(offs_n[:, None] < seqlen_k) & (offs_d[None, :] < headdim), other=0.0) - v = tl.load(v_ptrs, mask=(offs_n[:, None] < seqlen_k) & (offs_d[None, :] < headdim), other=0.0) - num_block_m = tl.cdiv(seqlen_q, BLOCK_M) - for start_m in range(begin_m, num_block_m * BLOCK_M, BLOCK_M): - start_m = tl.multiple_of(start_m, BLOCK_M) - offs_m_curr = start_m + offs_m - if EVEN_M & EVEN_HEADDIM: - q = tl.load(q_ptrs) - elif EVEN_HEADDIM: - q = tl.load(q_ptrs, mask=offs_m_curr[:, None] < seqlen_q, other=0.0) - else: - q = tl.load(q_ptrs, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0) - qk = tl.dot(q, k, trans_b=True) - if not EVEN_N: - qk = tl.where(offs_n[None, :] < seqlen_k, qk, float('-inf')) - if IS_CAUSAL: - qk = tl.where(offs_m_curr[:, None] >= offs_n[None, :], qk, float('-inf')) - if BIAS_TYPE != 'none': - tl.debug_barrier() - if BIAS_TYPE == 'vector': - if EVEN_N: - bias = tl.load(b_ptrs).to(tl.float32) - else: - bias = tl.load(b_ptrs, mask=offs_n < seqlen_k, other=0.0).to(tl.float32) - bias = bias[None, :] - elif BIAS_TYPE == 'matrix': - if EVEN_M & EVEN_N: - bias = tl.load(b_ptrs).to(tl.float32) - else: - bias = tl.load(b_ptrs, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_n[None, :] < seqlen_k), other=0.0).to(tl.float32) - qk = qk * softmax_scale + bias - if not EVEN_M & EVEN_HEADDIM: - tl.debug_barrier() - lse_i = tl.load(LSE + offs_m_curr) - if BIAS_TYPE == 'none': - p = tl.exp(qk * softmax_scale - lse_i[:, None]) - else: - p = tl.exp(qk - lse_i[:, None]) - if EVEN_M & EVEN_HEADDIM: - do = tl.load(do_ptrs) - else: - do = tl.load(do_ptrs, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0) - dv += tl.dot(p.to(do.dtype), do, trans_a=True) - if not EVEN_M & EVEN_HEADDIM: - tl.debug_barrier() - dp = tl.dot(do, v, trans_b=True) - if not EVEN_HEADDIM: - tl.debug_barrier() - Di = tl.load(D + offs_m_curr) - ds = (p * (dp - Di[:, None]) * softmax_scale).to(q.dtype) - dk += tl.dot(ds, q, trans_a=True) - if not EVEN_M & EVEN_HEADDIM: - tl.debug_barrier() - if not ATOMIC_ADD: - if EVEN_M & EVEN_HEADDIM: - dq = tl.load(dq_ptrs, eviction_policy='evict_last') - dq += tl.dot(ds, k) - tl.store(dq_ptrs, dq, eviction_policy='evict_last') - elif EVEN_HEADDIM: - dq = tl.load(dq_ptrs, mask=offs_m_curr[:, None] < seqlen_q, other=0.0, eviction_policy='evict_last') - dq += tl.dot(ds, k) - tl.store(dq_ptrs, dq, mask=offs_m_curr[:, None] < seqlen_q, eviction_policy='evict_last') - else: - dq = tl.load(dq_ptrs, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_d[None, :] < headdim), other=0.0, eviction_policy='evict_last') - dq += tl.dot(ds, k) - tl.store(dq_ptrs, dq, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_d[None, :] < headdim), eviction_policy='evict_last') - else: - dq = tl.dot(ds, k) - if EVEN_M & EVEN_HEADDIM: - tl.atomic_add(dq_ptrs, dq) - elif EVEN_HEADDIM: - tl.atomic_add(dq_ptrs, dq, mask=offs_m_curr[:, None] < seqlen_q) - else: - tl.atomic_add(dq_ptrs, dq, mask=(offs_m_curr[:, None] < seqlen_q) & (offs_d[None, :] < headdim)) - dq_ptrs += BLOCK_M * stride_dqm - q_ptrs += BLOCK_M * stride_qm - do_ptrs += BLOCK_M * stride_dom - if BIAS_TYPE == 'matrix': - b_ptrs += BLOCK_M * stride_bm - dv_ptrs = DV + (offs_n[:, None] * stride_dvn + offs_d[None, :]) - dk_ptrs = DK + (offs_n[:, None] * stride_dkn + offs_d[None, :]) - _bwd_store_dk_dv(dk_ptrs, dv_ptrs, dk, dv, offs_n, offs_d, seqlen_k, headdim, EVEN_M=EVEN_M, EVEN_N=EVEN_N, EVEN_HEADDIM=EVEN_HEADDIM) - -def init_to_zero(name): - return lambda nargs: nargs[name].zero_() - -@triton.autotune(configs=[triton.Config({'BLOCK_M': 128, 'BLOCK_N': 128, 'SEQUENCE_PARALLEL': False}, num_warps=8, num_stages=1, pre_hook=init_to_zero('DQ')), triton.Config({'BLOCK_M': 128, 'BLOCK_N': 128, 'SEQUENCE_PARALLEL': True}, num_warps=8, num_stages=1, pre_hook=init_to_zero('DQ'))], key=['CACHE_KEY_SEQLEN_Q', 'CACHE_KEY_SEQLEN_K', 'BIAS_TYPE', 'IS_CAUSAL', 'BLOCK_HEADDIM']) -@triton.heuristics({'EVEN_M': lambda args: args['seqlen_q'] % args['BLOCK_M'] == 0, 'EVEN_N': lambda args: args['seqlen_k'] % args['BLOCK_N'] == 0, 'EVEN_HEADDIM': lambda args: args['headdim'] == args['BLOCK_HEADDIM']}) -@triton.jit -def _bwd_kernel(Q, K, V, Bias, DO, DQ, DK, DV, LSE, D, softmax_scale, stride_qb, stride_qh, stride_qm, stride_kb, stride_kh, stride_kn, stride_vb, stride_vh, stride_vn, stride_bb, stride_bh, stride_bm, stride_dob, stride_doh, stride_dom, stride_dqb, stride_dqh, stride_dqm, stride_dkb, stride_dkh, stride_dkn, stride_dvb, stride_dvh, stride_dvn, nheads, seqlen_q, seqlen_k, seqlen_q_rounded, headdim, CACHE_KEY_SEQLEN_Q, CACHE_KEY_SEQLEN_K, BIAS_TYPE: tl.constexpr, IS_CAUSAL: tl.constexpr, BLOCK_HEADDIM: tl.constexpr, SEQUENCE_PARALLEL: tl.constexpr, EVEN_M: tl.constexpr, EVEN_N: tl.constexpr, EVEN_HEADDIM: tl.constexpr, BLOCK_M: tl.constexpr, BLOCK_N: tl.constexpr): - off_hb = tl.program_id(1) - off_b = off_hb // nheads - off_h = off_hb % nheads - Q += off_b * stride_qb + off_h * stride_qh - K += off_b * stride_kb + off_h * stride_kh - V += off_b * stride_vb + off_h * stride_vh - DO += off_b * stride_dob + off_h * stride_doh - DQ += off_b * stride_dqb + off_h * stride_dqh - DK += off_b * stride_dkb + off_h * stride_dkh - DV += off_b * stride_dvb + off_h * stride_dvh - if BIAS_TYPE != 'none': - Bias += off_b * stride_bb + off_h * stride_bh - D += off_hb * seqlen_q_rounded - LSE += off_hb * seqlen_q_rounded - if not SEQUENCE_PARALLEL: - num_block_n = tl.cdiv(seqlen_k, BLOCK_N) - for start_n in range(0, num_block_n): - _bwd_kernel_one_col_block(start_n, Q, K, V, Bias, DO, DQ, DK, DV, LSE, D, softmax_scale, stride_qm, stride_kn, stride_vn, stride_bm, stride_dom, stride_dqm, stride_dkn, stride_dvn, seqlen_q, seqlen_k, headdim, ATOMIC_ADD=False, BIAS_TYPE=BIAS_TYPE, IS_CAUSAL=IS_CAUSAL, BLOCK_HEADDIM=BLOCK_HEADDIM, EVEN_M=EVEN_M, EVEN_N=EVEN_N, EVEN_HEADDIM=EVEN_HEADDIM, BLOCK_M=BLOCK_M, BLOCK_N=BLOCK_N) - else: - start_n = tl.program_id(0) - _bwd_kernel_one_col_block(start_n, Q, K, V, Bias, DO, DQ, DK, DV, LSE, D, softmax_scale, stride_qm, stride_kn, stride_vn, stride_bm, stride_dom, stride_dqm, stride_dkn, stride_dvn, seqlen_q, seqlen_k, headdim, ATOMIC_ADD=True, BIAS_TYPE=BIAS_TYPE, IS_CAUSAL=IS_CAUSAL, BLOCK_HEADDIM=BLOCK_HEADDIM, EVEN_M=EVEN_M, EVEN_N=EVEN_N, EVEN_HEADDIM=EVEN_HEADDIM, BLOCK_M=BLOCK_M, BLOCK_N=BLOCK_N) - -def _flash_attn_forward(q, k, v, bias=None, causal=False, softmax_scale=None): - (batch, seqlen_q, nheads, d) = q.shape - (_, seqlen_k, _, _) = k.shape - assert k.shape == (batch, seqlen_k, nheads, d) - assert v.shape == (batch, seqlen_k, nheads, d) - assert d <= 128, 'FlashAttention only support head dimensions up to 128' - assert q.dtype == k.dtype == v.dtype, 'All tensors must have the same type' - assert q.dtype in [torch.float16, torch.bfloat16], 'Only support fp16 and bf16' - assert q.is_cuda and k.is_cuda and v.is_cuda - softmax_scale = softmax_scale or 1.0 / math.sqrt(d) - has_bias = bias is not None - bias_type = 'none' - if has_bias: - assert bias.dtype in [q.dtype, torch.float] - assert bias.is_cuda - assert bias.dim() == 4 - if bias.stride(-1) != 1: - bias = bias.contiguous() - if bias.shape[2:] == (1, seqlen_k): - bias_type = 'vector' - elif bias.shape[2:] == (seqlen_q, seqlen_k): - bias_type = 'matrix' - else: - raise RuntimeError('Last 2 dimensions of bias must be (1, seqlen_k) or (seqlen_q, seqlen_k)') - bias = bias.expand(batch, nheads, seqlen_q, seqlen_k) - bias_strides = (bias.stride(0), bias.stride(1), bias.stride(2)) if has_bias else (0, 0, 0) - seqlen_q_rounded = math.ceil(seqlen_q / 128) * 128 - lse = torch.empty((batch, nheads, seqlen_q_rounded), device=q.device, dtype=torch.float32) - tmp = torch.empty((batch, nheads, seqlen_q_rounded), device=q.device, dtype=torch.float32) - o = torch.empty_like(q) - BLOCK_HEADDIM = max(triton.next_power_of_2(d), 16) - BLOCK = 128 - num_warps = 4 if d <= 64 else 8 - grid = lambda META: (triton.cdiv(seqlen_q, META['BLOCK_M']), batch * nheads) - _fwd_kernel[grid](q, k, v, bias, o, lse, tmp, softmax_scale, q.stride(0), q.stride(2), q.stride(1), k.stride(0), k.stride(2), k.stride(1), v.stride(0), v.stride(2), v.stride(1), *bias_strides, o.stride(0), o.stride(2), o.stride(1), nheads, seqlen_q, seqlen_k, seqlen_q_rounded, d, seqlen_q // 32, seqlen_k // 32, bias_type, causal, BLOCK_HEADDIM, BLOCK_M=BLOCK, BLOCK_N=BLOCK, num_warps=num_warps, num_stages=1) - return (o, lse, softmax_scale) - -def _flash_attn_backward(do, q, k, v, o, lse, dq, dk, dv, bias=None, causal=False, softmax_scale=None): - if do.stride(-1) != 1: - do = do.contiguous() - (batch, seqlen_q, nheads, d) = q.shape - (_, seqlen_k, _, _) = k.shape - assert d <= 128 - seqlen_q_rounded = math.ceil(seqlen_q / 128) * 128 - assert lse.shape == (batch, nheads, seqlen_q_rounded) - assert q.stride(-1) == k.stride(-1) == v.stride(-1) == o.stride(-1) == 1 - assert dq.stride(-1) == dk.stride(-1) == dv.stride(-1) == 1 - softmax_scale = softmax_scale or 1.0 / math.sqrt(d) - dq_accum = torch.empty_like(q, dtype=torch.float32) - delta = torch.empty_like(lse) - BLOCK_HEADDIM = max(triton.next_power_of_2(d), 16) - grid = lambda META: (triton.cdiv(seqlen_q, META['BLOCK_M']), batch * nheads) - _bwd_preprocess_do_o_dot[grid](o, do, delta, o.stride(0), o.stride(2), o.stride(1), do.stride(0), do.stride(2), do.stride(1), nheads, seqlen_q, seqlen_q_rounded, d, BLOCK_M=128, BLOCK_HEADDIM=BLOCK_HEADDIM) - has_bias = bias is not None - bias_type = 'none' - if has_bias: - assert bias.dtype in [q.dtype, torch.float] - assert bias.is_cuda - assert bias.dim() == 4 - assert bias.stride(-1) == 1 - if bias.shape[2:] == (1, seqlen_k): - bias_type = 'vector' - elif bias.shape[2:] == (seqlen_q, seqlen_k): - bias_type = 'matrix' - else: - raise RuntimeError('Last 2 dimensions of bias must be (1, seqlen_k) or (seqlen_q, seqlen_k)') - bias = bias.expand(batch, nheads, seqlen_q, seqlen_k) - bias_strides = (bias.stride(0), bias.stride(1), bias.stride(2)) if has_bias else (0, 0, 0) - grid = lambda META: (triton.cdiv(seqlen_k, META['BLOCK_N']) if META['SEQUENCE_PARALLEL'] else 1, batch * nheads) - _bwd_kernel[grid](q, k, v, bias, do, dq_accum, dk, dv, lse, delta, softmax_scale, q.stride(0), q.stride(2), q.stride(1), k.stride(0), k.stride(2), k.stride(1), v.stride(0), v.stride(2), v.stride(1), *bias_strides, do.stride(0), do.stride(2), do.stride(1), dq_accum.stride(0), dq_accum.stride(2), dq_accum.stride(1), dk.stride(0), dk.stride(2), dk.stride(1), dv.stride(0), dv.stride(2), dv.stride(1), nheads, seqlen_q, seqlen_k, seqlen_q_rounded, d, seqlen_q // 32, seqlen_k // 32, bias_type, causal, BLOCK_HEADDIM) - dq.copy_(dq_accum) - -class FlashAttnQKVPackedFunc(torch.autograd.Function): - - @staticmethod - def forward(ctx, qkv, bias=None, causal=False, softmax_scale=None): - """ - qkv: (batch, seqlen, 3, nheads, headdim) - bias: optional, shape broadcastible to (batch, nheads, seqlen, seqlen). - For example, ALiBi mask for causal would have shape (1, nheads, 1, seqlen). - ALiBi mask for non-causal would have shape (1, nheads, seqlen, seqlen) - """ - if qkv.stride(-1) != 1: - qkv = qkv.contiguous() - (o, lse, ctx.softmax_scale) = _flash_attn_forward(qkv[:, :, 0], qkv[:, :, 1], qkv[:, :, 2], bias=bias, causal=causal, softmax_scale=softmax_scale) - ctx.save_for_backward(qkv, o, lse, bias) - ctx.causal = causal - return o - - @staticmethod - def backward(ctx, do): - (qkv, o, lse, bias) = ctx.saved_tensors - assert not ctx.needs_input_grad[1], 'FlashAttention does not support bias gradient yet' - with torch.inference_mode(): - dqkv = torch.empty_like(qkv) - _flash_attn_backward(do, qkv[:, :, 0], qkv[:, :, 1], qkv[:, :, 2], o, lse, dqkv[:, :, 0], dqkv[:, :, 1], dqkv[:, :, 2], bias=bias, causal=ctx.causal, softmax_scale=ctx.softmax_scale) - return (dqkv, None, None, None) -flash_attn_qkvpacked_func = FlashAttnQKVPackedFunc.apply - -class FlashAttnKVPackedFunc(torch.autograd.Function): - - @staticmethod - def forward(ctx, q, kv, bias=None, causal=False, softmax_scale=None): - """ - q: (batch, seqlen_q, nheads, headdim) - kv: (batch, seqlen_k, 2, nheads, headdim) - bias: optional, shape broadcastible to (batch, nheads, seqlen_q, seqlen_k). - For example, ALiBi mask for causal would have shape (1, nheads, 1, seqlen_k). - ALiBi mask for non-causal would have shape (1, nheads, seqlen_q, seqlen_k) - """ - (q, kv) = [x if x.stride(-1) == 1 else x.contiguous() for x in [q, kv]] - (o, lse, ctx.softmax_scale) = _flash_attn_forward(q, kv[:, :, 0], kv[:, :, 1], bias=bias, causal=causal, softmax_scale=softmax_scale) - ctx.save_for_backward(q, kv, o, lse, bias) - ctx.causal = causal - return o - - @staticmethod - def backward(ctx, do): - (q, kv, o, lse, bias) = ctx.saved_tensors - if len(ctx.needs_input_grad) >= 3: - assert not ctx.needs_input_grad[2], 'FlashAttention does not support bias gradient yet' - with torch.inference_mode(): - dq = torch.empty_like(q) - dkv = torch.empty_like(kv) - _flash_attn_backward(do, q, kv[:, :, 0], kv[:, :, 1], o, lse, dq, dkv[:, :, 0], dkv[:, :, 1], bias=bias, causal=ctx.causal, softmax_scale=ctx.softmax_scale) - return (dq, dkv, None, None, None) -flash_attn_kvpacked_func = FlashAttnKVPackedFunc.apply - -class FlashAttnFunc(torch.autograd.Function): - - @staticmethod - def forward(ctx, q, k, v, bias=None, causal=False, softmax_scale=None): - """ - q: (batch_size, seqlen_q, nheads, headdim) - k, v: (batch_size, seqlen_k, nheads, headdim) - bias: optional, shape broadcastible to (batch, nheads, seqlen_q, seqlen_k). - For example, ALiBi mask for causal would have shape (1, nheads, 1, seqlen_k). - ALiBi mask for non-causal would have shape (1, nheads, seqlen_q, seqlen_k) - """ - (q, k, v) = [x if x.stride(-1) == 1 else x.contiguous() for x in [q, k, v]] - (o, lse, ctx.softmax_scale) = _flash_attn_forward(q, k, v, bias=bias, causal=causal, softmax_scale=softmax_scale) - ctx.save_for_backward(q, k, v, o, lse, bias) - ctx.causal = causal - return o - - @staticmethod - def backward(ctx, do): - (q, k, v, o, lse, bias) = ctx.saved_tensors - assert not ctx.needs_input_grad[3], 'FlashAttention does not support bias gradient yet' - with torch.inference_mode(): - dq = torch.empty_like(q) - dk = torch.empty_like(k) - dv = torch.empty_like(v) - _flash_attn_backward(do, q, k, v, o, lse, dq, dk, dv, bias=bias, causal=ctx.causal, softmax_scale=ctx.softmax_scale) - return (dq, dk, dv, None, None, None) -flash_attn_func = FlashAttnFunc.apply \ No newline at end of file diff --git a/llava/model/language_model/mpt/hf_prefixlm_converter.py b/llava/model/language_model/mpt/hf_prefixlm_converter.py deleted file mode 100644 index 8c1a64872..000000000 --- a/llava/model/language_model/mpt/hf_prefixlm_converter.py +++ /dev/null @@ -1,415 +0,0 @@ -"""Converts Huggingface Causal LM to Prefix LM. - -Conversion does lightweight surgery on a HuggingFace -Causal LM to convert it to a Prefix LM. - -Prefix LMs accepts a `bidirectional_mask` input in `forward` -and treat the input prompt as the prefix in `generate`. -""" -import math -import warnings -from types import MethodType -from typing import Any, Dict, List, Optional, Tuple, Union -import torch -from transformers.models.bloom.modeling_bloom import BaseModelOutputWithPastAndCrossAttentions, BloomForCausalLM, BloomModel, CausalLMOutputWithCrossAttentions, CrossEntropyLoss -from transformers.models.bloom.modeling_bloom import _expand_mask as _expand_mask_bloom -from transformers.models.bloom.modeling_bloom import _make_causal_mask as _make_causal_mask_bloom -from transformers.models.bloom.modeling_bloom import logging -from transformers.models.gpt2.modeling_gpt2 import GPT2LMHeadModel -from transformers.models.gpt_neo.modeling_gpt_neo import GPTNeoForCausalLM -from transformers.models.gpt_neox.modeling_gpt_neox import GPTNeoXForCausalLM -from transformers.models.gptj.modeling_gptj import GPTJForCausalLM -from transformers.models.opt.modeling_opt import OPTForCausalLM -from transformers.models.opt.modeling_opt import _expand_mask as _expand_mask_opt -from transformers.models.opt.modeling_opt import _make_causal_mask as _make_causal_mask_opt -logger = logging.get_logger(__name__) -_SUPPORTED_GPT_MODELS = (GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPTNeoXForCausalLM) -CAUSAL_GPT_TYPES = Union[GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPTNeoXForCausalLM] - -def _convert_gpt_causal_lm_to_prefix_lm(model: CAUSAL_GPT_TYPES) -> CAUSAL_GPT_TYPES: - """Converts a GPT-style Causal LM to a Prefix LM. - - Supported HuggingFace model classes: - - `GPT2LMHeadModel` - - `GPTNeoForCausalLM` - - `GPTNeoXForCausalLM` - - `GPTJForCausalLM` - - See `convert_hf_causal_lm_to_prefix_lm` for more details. - """ - if hasattr(model, '_prefix_lm_converted'): - return model - assert isinstance(model, _SUPPORTED_GPT_MODELS) - assert model.config.add_cross_attention == False, 'Only supports GPT-style decoder-only models' - - def _get_attn_modules(model: CAUSAL_GPT_TYPES) -> List[torch.nn.Module]: - """Helper that gets a list of the model's attention modules. - - Each module has a `bias` buffer used for causal masking. The Prefix LM - conversion adds logic to dynamically manipulate these biases to support - Prefix LM attention masking. - """ - attn_modules = [] - if isinstance(model, GPTNeoXForCausalLM): - blocks = model.gpt_neox.layers - else: - blocks = model.transformer.h - for block in blocks: - if isinstance(model, GPTNeoForCausalLM): - if block.attn.attention_type != 'global': - continue - attn_module = block.attn.attention - elif isinstance(model, GPTNeoXForCausalLM): - attn_module = block.attention - else: - attn_module = block.attn - attn_modules.append(attn_module) - return attn_modules - setattr(model, '_original_forward', getattr(model, 'forward')) - setattr(model, '_original_generate', getattr(model, 'generate')) - - def forward(self: CAUSAL_GPT_TYPES, input_ids: Optional[torch.LongTensor]=None, past_key_values: Optional[Tuple[Tuple[torch.Tensor]]]=None, attention_mask: Optional[torch.FloatTensor]=None, bidirectional_mask: Optional[torch.Tensor]=None, token_type_ids: Optional[torch.LongTensor]=None, position_ids: Optional[torch.LongTensor]=None, head_mask: Optional[torch.FloatTensor]=None, inputs_embeds: Optional[torch.FloatTensor]=None, labels: Optional[torch.LongTensor]=None, use_cache: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, return_dict: Optional[bool]=None): - """Wraps original forward to enable PrefixLM attention.""" - - def call_og_forward(): - if isinstance(self, GPTNeoXForCausalLM): - return self._original_forward(input_ids=input_ids, past_key_values=past_key_values, attention_mask=attention_mask, head_mask=head_mask, inputs_embeds=inputs_embeds, labels=labels, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict) - else: - return self._original_forward(input_ids=input_ids, past_key_values=past_key_values, attention_mask=attention_mask, token_type_ids=token_type_ids, position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, labels=labels, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict) - if bidirectional_mask is None: - return call_og_forward() - assert isinstance(bidirectional_mask, torch.Tensor) - attn_modules = _get_attn_modules(model) - (b, s) = bidirectional_mask.shape - max_length = attn_modules[0].bias.shape[-1] - if s > max_length: - raise ValueError(f'bidirectional_mask sequence length (={s}) exceeds the ' + f'max length allowed by the model ({max_length}).') - assert s <= max_length - if s < max_length: - pad = torch.zeros((int(b), int(max_length - s)), dtype=bidirectional_mask.dtype, device=bidirectional_mask.device) - bidirectional_mask = torch.cat([bidirectional_mask, pad], dim=1) - bidirectional = bidirectional_mask.unsqueeze(1).unsqueeze(1) - for attn_module in attn_modules: - attn_module.bias.data = torch.logical_or(attn_module.bias.data, bidirectional) - output = call_og_forward() - for attn_module in attn_modules: - attn_module.bias.data = torch.tril(attn_module.bias.data[0, 0])[None, None] - return output - - def generate(self: CAUSAL_GPT_TYPES, *args: tuple, **kwargs: Dict[str, Any]): - """Wraps original generate to enable PrefixLM attention.""" - attn_modules = _get_attn_modules(model) - for attn_module in attn_modules: - attn_module.bias.data[:] = 1 - output = self._original_generate(*args, **kwargs) - for attn_module in attn_modules: - attn_module.bias.data = torch.tril(attn_module.bias.data[0, 0])[None, None] - return output - setattr(model, 'forward', MethodType(forward, model)) - setattr(model, 'generate', MethodType(generate, model)) - setattr(model, '_prefix_lm_converted', True) - return model - -def _convert_bloom_causal_lm_to_prefix_lm(model: BloomForCausalLM) -> BloomForCausalLM: - """Converts a BLOOM Causal LM to a Prefix LM. - - Supported HuggingFace model classes: - - `BloomForCausalLM` - - See `convert_hf_causal_lm_to_prefix_lm` for more details. - """ - if hasattr(model, '_prefix_lm_converted'): - return model - assert isinstance(model, BloomForCausalLM) - assert model.config.add_cross_attention == False, 'Only supports BLOOM decoder-only models' - - def _prepare_attn_mask(self: BloomModel, attention_mask: torch.Tensor, bidirectional_mask: Optional[torch.Tensor], input_shape: Tuple[int, int], past_key_values_length: int) -> torch.BoolTensor: - combined_attention_mask = None - device = attention_mask.device - (_, src_length) = input_shape - if src_length > 1: - combined_attention_mask = _make_causal_mask_bloom(input_shape, device=device, past_key_values_length=past_key_values_length) - if bidirectional_mask is not None: - assert attention_mask.shape == bidirectional_mask.shape - expanded_bidirectional_mask = _expand_mask_bloom(bidirectional_mask, tgt_length=src_length) - combined_attention_mask = torch.logical_and(combined_attention_mask, expanded_bidirectional_mask) - expanded_attn_mask = _expand_mask_bloom(attention_mask, tgt_length=src_length) - combined_attention_mask = expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask | combined_attention_mask - return combined_attention_mask - - def _build_alibi_tensor(self: BloomModel, batch_size: int, query_length: int, key_length: int, dtype: torch.dtype, device: torch.device) -> torch.Tensor: - num_heads = self.config.n_head - closest_power_of_2 = 2 ** math.floor(math.log2(num_heads)) - base = torch.tensor(2 ** (-2 ** (-(math.log2(closest_power_of_2) - 3))), device=device, dtype=torch.float32) - powers = torch.arange(1, 1 + closest_power_of_2, device=device, dtype=torch.int32) - slopes = torch.pow(base, powers) - if closest_power_of_2 != num_heads: - extra_base = torch.tensor(2 ** (-2 ** (-(math.log2(2 * closest_power_of_2) - 3))), device=device, dtype=torch.float32) - num_remaining_heads = min(closest_power_of_2, num_heads - closest_power_of_2) - extra_powers = torch.arange(1, 1 + 2 * num_remaining_heads, 2, device=device, dtype=torch.int32) - slopes = torch.cat([slopes, torch.pow(extra_base, extra_powers)], dim=0) - qa = torch.arange(query_length, device=device, dtype=torch.int32).view(-1, 1) - ka = torch.arange(key_length, device=device, dtype=torch.int32).view(1, -1) - diffs = qa - ka + key_length - query_length - diffs = -diffs.abs() - alibi = slopes.view(1, num_heads, 1, 1) * diffs.view(1, 1, query_length, key_length) - alibi = alibi.expand(batch_size, -1, -1, -1).reshape(-1, query_length, key_length) - return alibi.to(dtype) - KeyValueT = Tuple[torch.Tensor, torch.Tensor] - - def forward(self: BloomModel, input_ids: Optional[torch.LongTensor]=None, past_key_values: Optional[Tuple[KeyValueT, ...]]=None, attention_mask: Optional[torch.Tensor]=None, bidirectional_mask: Optional[torch.Tensor]=None, head_mask: Optional[torch.LongTensor]=None, inputs_embeds: Optional[torch.LongTensor]=None, use_cache: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, return_dict: Optional[bool]=None, **deprecated_arguments) -> Union[Tuple[torch.Tensor, ...], BaseModelOutputWithPastAndCrossAttentions]: - if deprecated_arguments.pop('position_ids', False) is not False: - warnings.warn('`position_ids` have no functionality in BLOOM and will be removed in v5.0.0. ' + 'You can safely ignore passing `position_ids`.', FutureWarning) - if len(deprecated_arguments) > 0: - raise ValueError(f'Got unexpected arguments: {deprecated_arguments}') - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - use_cache = use_cache if use_cache is not None else self.config.use_cache - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if input_ids is not None and inputs_embeds is not None: - raise ValueError('You cannot specify both input_ids and inputs_embeds at the same time') - elif input_ids is not None: - (batch_size, seq_length) = input_ids.shape - elif inputs_embeds is not None: - (batch_size, seq_length, _) = inputs_embeds.shape - else: - raise ValueError('You have to specify either input_ids or inputs_embeds') - if past_key_values is None: - past_key_values = tuple([None] * len(self.h)) - head_mask = self.get_head_mask(head_mask, self.config.n_layer) - if inputs_embeds is None: - inputs_embeds = self.word_embeddings(input_ids) - hidden_states = self.word_embeddings_layernorm(inputs_embeds) - presents = () if use_cache else None - all_self_attentions = () if output_attentions else None - all_hidden_states = () if output_hidden_states else None - seq_length_with_past = seq_length - past_key_values_length = 0 - if past_key_values[0] is not None: - tmp = past_key_values[0][0] - past_key_values_length = tmp.shape[2] - seq_length_with_past = seq_length_with_past + past_key_values_length - if attention_mask is None: - attention_mask = torch.ones((batch_size, seq_length_with_past), device=hidden_states.device) - else: - attention_mask = attention_mask.to(hidden_states.device) - alibi = self._build_alibi_tensor(batch_size=batch_size, query_length=seq_length, key_length=seq_length_with_past, dtype=hidden_states.dtype, device=hidden_states.device) - causal_mask = self._prepare_attn_mask(attention_mask, bidirectional_mask, input_shape=(batch_size, seq_length), past_key_values_length=past_key_values_length) - for (i, (block, layer_past)) in enumerate(zip(self.h, past_key_values)): - if output_hidden_states: - hst = (hidden_states,) - all_hidden_states = all_hidden_states + hst - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning('`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...') - use_cache = False - - def create_custom_forward(module): - - def custom_forward(*inputs): - return module(*inputs, use_cache=use_cache, output_attentions=output_attentions) - return custom_forward - outputs = torch.utils.checkpoint.checkpoint(create_custom_forward(block), hidden_states, alibi, causal_mask, head_mask[i]) - else: - outputs = block(hidden_states, layer_past=layer_past, attention_mask=causal_mask, head_mask=head_mask[i], use_cache=use_cache, output_attentions=output_attentions, alibi=alibi) - hidden_states = outputs[0] - if use_cache is True: - presents = presents + (outputs[1],) - if output_attentions: - oa = (outputs[2 if use_cache else 1],) - all_self_attentions = all_self_attentions + oa - hidden_states = self.ln_f(hidden_states) - if output_hidden_states: - hst = (hidden_states,) - all_hidden_states = all_hidden_states + hst - if not return_dict: - return tuple((v for v in [hidden_states, presents, all_hidden_states, all_self_attentions] if v is not None)) - return BaseModelOutputWithPastAndCrossAttentions(last_hidden_state=hidden_states, past_key_values=presents, hidden_states=all_hidden_states, attentions=all_self_attentions) - setattr(model.transformer, '_prepare_attn_mask', MethodType(_prepare_attn_mask, model.transformer)) - setattr(model.transformer, '_build_alibi_tensor', MethodType(_build_alibi_tensor, model.transformer)) - setattr(model.transformer, 'forward', MethodType(forward, model.transformer)) - KeyValueT = Tuple[torch.Tensor, torch.Tensor] - - def forward(self: BloomForCausalLM, input_ids: Optional[torch.LongTensor]=None, past_key_values: Optional[Tuple[KeyValueT, ...]]=None, attention_mask: Optional[torch.Tensor]=None, bidirectional_mask: Optional[torch.Tensor]=None, head_mask: Optional[torch.Tensor]=None, inputs_embeds: Optional[torch.Tensor]=None, labels: Optional[torch.Tensor]=None, use_cache: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, return_dict: Optional[bool]=None, **deprecated_arguments) -> Union[Tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: - """Replacement forward method for BloomCausalLM.""" - if deprecated_arguments.pop('position_ids', False) is not False: - warnings.warn('`position_ids` have no functionality in BLOOM and will be removed ' + 'in v5.0.0. You can safely ignore passing `position_ids`.', FutureWarning) - if len(deprecated_arguments) > 0: - raise ValueError(f'Got unexpected arguments: {deprecated_arguments}') - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - transformer_outputs = self.transformer(input_ids, past_key_values=past_key_values, attention_mask=attention_mask, bidirectional_mask=bidirectional_mask, head_mask=head_mask, inputs_embeds=inputs_embeds, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict) - hidden_states = transformer_outputs[0] - lm_logits = self.lm_head(hidden_states) - loss = None - if labels is not None: - shift_logits = lm_logits[..., :-1, :].contiguous() - shift_labels = labels[..., 1:].contiguous() - (batch_size, seq_length, vocab_size) = shift_logits.shape - loss_fct = CrossEntropyLoss() - loss = loss_fct(shift_logits.view(batch_size * seq_length, vocab_size), shift_labels.view(batch_size * seq_length)) - if not return_dict: - output = (lm_logits,) + transformer_outputs[1:] - return (loss,) + output if loss is not None else output - return CausalLMOutputWithCrossAttentions(loss=loss, logits=lm_logits, past_key_values=transformer_outputs.past_key_values, hidden_states=transformer_outputs.hidden_states, attentions=transformer_outputs.attentions) - - def prepare_inputs_for_generation(self: BloomForCausalLM, input_ids: torch.LongTensor, past: Optional[torch.Tensor]=None, attention_mask: Optional[torch.Tensor]=None, **kwargs) -> dict: - if past: - input_ids = input_ids[:, -1].unsqueeze(-1) - bidirectional_mask = None - if past[0][0].shape[0] == input_ids.shape[0]: - past = self._convert_to_bloom_cache(past) - else: - bidirectional_mask = torch.ones_like(input_ids) - return {'input_ids': input_ids, 'past_key_values': past, 'use_cache': True, 'attention_mask': attention_mask, 'bidirectional_mask': bidirectional_mask} - setattr(model, 'forward', MethodType(forward, model)) - setattr(model, 'prepare_inputs_for_generation', MethodType(prepare_inputs_for_generation, model)) - setattr(model, '_prefix_lm_converted', True) - return model - -def _convert_opt_causal_lm_to_prefix_lm(model: OPTForCausalLM) -> OPTForCausalLM: - """Converts an OPT Causal LM to a Prefix LM. - - Supported HuggingFace model classes: - - `OPTForCausalLM` - - See `convert_hf_causal_lm_to_prefix_lm` for more details. - """ - if hasattr(model, '_prefix_lm_converted'): - return model - assert isinstance(model, OPTForCausalLM) - assert model.config.add_cross_attention == False, 'Only supports OPT decoder-only models' - setattr(model, '_original_forward', getattr(model, 'forward')) - setattr(model, '_original_generate', getattr(model, 'generate')) - model.model.decoder.bidirectional_mask = None - - def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length): - combined_attention_mask = None - if input_shape[-1] > 1: - if self.bidirectional_mask == 'g': - (bsz, src_length) = input_shape - combined_attention_mask = torch.zeros((bsz, 1, src_length, src_length + past_key_values_length), dtype=inputs_embeds.dtype, device=inputs_embeds.device) - else: - combined_attention_mask = _make_causal_mask_opt(input_shape, inputs_embeds.dtype, past_key_values_length=past_key_values_length).to(inputs_embeds.device) - if self.bidirectional_mask is not None: - assert attention_mask.shape == self.bidirectional_mask.shape - expanded_bidirectional_mask = _expand_mask_opt(self.bidirectional_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(inputs_embeds.device) - combined_attention_mask = torch.maximum(expanded_bidirectional_mask, combined_attention_mask) - if attention_mask is not None: - expanded_attn_mask = _expand_mask_opt(attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(inputs_embeds.device) - combined_attention_mask = expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask + combined_attention_mask - return combined_attention_mask - setattr(model.model.decoder, '_prepare_decoder_attention_mask', MethodType(_prepare_decoder_attention_mask, model.model.decoder)) - - def forward(self: OPTForCausalLM, input_ids: Optional[torch.LongTensor]=None, attention_mask: Optional[torch.Tensor]=None, bidirectional_mask: Optional[torch.ByteTensor]=None, head_mask: Optional[torch.Tensor]=None, past_key_values: Optional[List[torch.FloatTensor]]=None, inputs_embeds: Optional[torch.FloatTensor]=None, labels: Optional[torch.LongTensor]=None, use_cache: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, return_dict: Optional[bool]=None): - - def call_og_forward(): - return self._original_forward(input_ids=input_ids, attention_mask=attention_mask, head_mask=head_mask, past_key_values=past_key_values, inputs_embeds=inputs_embeds, labels=labels, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict) - if bidirectional_mask is None: - return call_og_forward() - self.model.decoder.bidirectional_mask = bidirectional_mask - try: - outputs = call_og_forward() - except: - self.model.decoder.bidirectional_mask = None - raise - self.model.decoder.bidirectional_mask = None - return outputs - - def generate(self: OPTForCausalLM, *args: tuple, **kwargs: Dict[str, Any]): - """Wraps original generate to enable PrefixLM-style attention.""" - self.model.decoder.bidirectional_mask = 'g' - try: - output = self._original_generate(*args, **kwargs) - except: - self.model.decoder.bidirectional_mask = None - raise - self.model.decoder.bidirectional_mask = None - return output - setattr(model, 'forward', MethodType(forward, model)) - setattr(model, 'generate', MethodType(generate, model)) - setattr(model, '_prefix_lm_converted', True) - return model -_SUPPORTED_HF_MODELS = _SUPPORTED_GPT_MODELS + (BloomForCausalLM, OPTForCausalLM) -CAUSAL_LM_TYPES = Union[GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPTNeoXForCausalLM, BloomForCausalLM, OPTForCausalLM] - -def convert_hf_causal_lm_to_prefix_lm(model: CAUSAL_LM_TYPES) -> CAUSAL_LM_TYPES: - """Converts a HuggingFace Causal LM to a Prefix LM. - - Supported HuggingFace model classes: - - `GPT2LMHeadModel` - - `GPTNeoForCausalLM` - - `GPTNeoXForCausalLM` - - `GPTJForCausalLM` - - `BloomForCausalLM` - - `OPTForCausalLM` - - Conversion to a Prefix LM is done by modifying the `forward` method, and possibly also the - `generate` method and/or select underlying methods depending on the model class. - - These changes preserve the model API, but add a new input to `forward`: "bidirectional_mask". - - Notes on training: - To actually train the converted model as a Prefix LM, training batches will need to indicate - the prefix/target structure by including `bidirectional_mask` as part of the batch inputs. - - **This is not a standard input and requires custom layers either within or after your dataloader.** - - In addition to adding `bidirectional_mask` to the batch, this custom code should modify `labels` - such that `batch['labels'][batch['bidirectional_mask'] == 1] == -100`. - That is, the prefix portion of the sequence should not generate any loss. Loss should only be - generated by the target portion of the sequence. - - Notes on `GPTNeoForCausalLM`: - To simplify the implementation, "global" and "local" attention layers are handled differently. - For "global" layers, we handle conversion as described above. For "local" layers, which use a - causal attention mask within a restricted local window, we do not alter the masking. - - Notes on `forward` method conversion: - After conversion, the `forward` method will handle a new input, `bidirectional_mask`, - which should be a [batch_size, seq_length] byte tensor, where 1 indicates token positions - belonging to the prefix (prefix tokens can attend to one another bidirectionally), and - 0 indicates token positions belonging to the target. - - The new `forward` method will incorporate `bidirectional_mask` (if supplied) into the existing - causal mask, call the original `forward` method, and (if the causal mask is a buffer) reset - the causal masks before returning the result. - - Notes on `generate` method conversion: - After conversion, the `generate` method will have the same signature but will internally - convert all causal masks to be purely bidirectional, call the original `generate` method, and - (where appropriate) reset the causal masks before returning the result. - - This works thanks to the logic of the HuggingFace `generate` API, which first encodes the token - "prompt" passed to `generate` (which is treated as the prefix) and then sequentially generates - each new token. Encodings are cached as generation happens, so all prefix tokens can attend to one - another (as expected in a Prefix LM) and generated tokens can only attend to prefix tokens and - previously-generated tokens (also as expected in a Prefix LM). - - To preserve the API, the original methods are renamed to `_original_forward` and - `_original_generate`, and replaced with new `forward` and `generate` methods that wrap - them, respectively. Although implementation details vary by model class. - """ - if isinstance(model, _SUPPORTED_GPT_MODELS): - return _convert_gpt_causal_lm_to_prefix_lm(model) - elif isinstance(model, BloomForCausalLM): - return _convert_bloom_causal_lm_to_prefix_lm(model) - elif isinstance(model, OPTForCausalLM): - return _convert_opt_causal_lm_to_prefix_lm(model) - else: - raise TypeError(f'Cannot convert model to Prefix LM. ' + f'Model does not belong to set of supported HF models:' + f'\n{_SUPPORTED_HF_MODELS}') - -def add_bidirectional_mask_if_missing(batch: Dict[str, Any]): - """Attempts to add bidirectional_mask to batch if missing. - - Raises: - KeyError if bidirectional_mask is missing and can't be inferred - """ - if 'bidirectional_mask' not in batch: - if batch.get('mode', None) == 'icl_task': - batch['bidirectional_mask'] = batch['attention_mask'].clone() - for (i, continuation_indices) in enumerate(batch['continuation_indices']): - batch['bidirectional_mask'][i, continuation_indices] = 0 - elif 'labels' in batch and 'attention_mask' in batch: - batch['bidirectional_mask'] = torch.logical_and(torch.eq(batch['attention_mask'], 1), torch.eq(batch['labels'], -100)).type_as(batch['attention_mask']) - else: - raise KeyError('No bidirectional_mask in batch and not sure how to construct one.') \ No newline at end of file diff --git a/llava/model/language_model/mpt/meta_init_context.py b/llava/model/language_model/mpt/meta_init_context.py deleted file mode 100644 index 6cba6fff0..000000000 --- a/llava/model/language_model/mpt/meta_init_context.py +++ /dev/null @@ -1,94 +0,0 @@ -from contextlib import contextmanager -import torch -import torch.nn as nn - -@contextmanager -def init_empty_weights(include_buffers: bool=False): - """Meta initialization context manager. - - A context manager under which models are initialized with all parameters - on the meta device, therefore creating an empty model. Useful when just - initializing the model would blow the available RAM. - - Args: - include_buffers (`bool`, *optional*, defaults to `False`): Whether or - not to also put all buffers on the meta device while initializing. - - Example: - ```python - import torch.nn as nn - - # Initialize a model with 100 billions parameters in no time and without using any RAM. - with init_empty_weights(): - tst = nn.Sequential(*[nn.Linear(10000, 10000) for _ in range(1000)]) - ``` - - - - Any model created under this context manager has no weights. As such you can't do something like - `model.to(some_device)` with it. To load weights inside your empty model, see [`load_checkpoint_and_dispatch`]. - - - """ - with init_on_device(torch.device('meta'), include_buffers=include_buffers) as f: - yield f - -@contextmanager -def init_on_device(device: torch.device, include_buffers: bool=False): - """Device initialization context manager. - - A context manager under which models are initialized with all parameters - on the specified device. - - Args: - device (`torch.device`): Device to initialize all parameters on. - include_buffers (`bool`, *optional*, defaults to `False`): Whether or - not to also put all buffers on the meta device while initializing. - - Example: - ```python - import torch.nn as nn - - with init_on_device(device=torch.device("cuda")): - tst = nn.Liner(100, 100) # on `cuda` device - ``` - """ - old_register_parameter = nn.Module.register_parameter - if include_buffers: - old_register_buffer = nn.Module.register_buffer - - def register_empty_parameter(module, name, param): - old_register_parameter(module, name, param) - if param is not None: - param_cls = type(module._parameters[name]) - kwargs = module._parameters[name].__dict__ - module._parameters[name] = param_cls(module._parameters[name].to(device), **kwargs) - - def register_empty_buffer(module, name, buffer): - old_register_buffer(module, name, buffer) - if buffer is not None: - module._buffers[name] = module._buffers[name].to(device) - if include_buffers: - tensor_constructors_to_patch = {torch_function_name: getattr(torch, torch_function_name) for torch_function_name in ['empty', 'zeros', 'ones', 'full']} - else: - tensor_constructors_to_patch = {} - - def patch_tensor_constructor(fn): - - def wrapper(*args, **kwargs): - kwargs['device'] = device - return fn(*args, **kwargs) - return wrapper - try: - nn.Module.register_parameter = register_empty_parameter - if include_buffers: - nn.Module.register_buffer = register_empty_buffer - for torch_function_name in tensor_constructors_to_patch.keys(): - setattr(torch, torch_function_name, patch_tensor_constructor(getattr(torch, torch_function_name))) - yield - finally: - nn.Module.register_parameter = old_register_parameter - if include_buffers: - nn.Module.register_buffer = old_register_buffer - for (torch_function_name, old_torch_function) in tensor_constructors_to_patch.items(): - setattr(torch, torch_function_name, old_torch_function) \ No newline at end of file diff --git a/llava/model/language_model/mpt/modeling_mpt.py b/llava/model/language_model/mpt/modeling_mpt.py deleted file mode 100644 index 13313441b..000000000 --- a/llava/model/language_model/mpt/modeling_mpt.py +++ /dev/null @@ -1,331 +0,0 @@ -"""A simple, flexible implementation of a GPT model. - -Inspired by https://github.com/karpathy/minGPT/blob/master/mingpt/model.py -""" -import math -import warnings -from typing import List, Optional, Tuple, Union -import torch -import torch.nn as nn -import torch.nn.functional as F -from transformers import PreTrainedModel, PreTrainedTokenizer, PreTrainedTokenizerFast -from transformers.modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast -from .attention import attn_bias_shape, build_attn_bias -from .blocks import MPTBlock -from .custom_embedding import SharedEmbedding -from .norm import NORM_CLASS_REGISTRY -from .configuration_mpt import MPTConfig -from .adapt_tokenizer import AutoTokenizerForMOD, adapt_tokenizer_for_denoising -from .hf_prefixlm_converter import add_bidirectional_mask_if_missing, convert_hf_causal_lm_to_prefix_lm -from .meta_init_context import init_empty_weights -from .param_init_fns import MODEL_INIT_REGISTRY, generic_param_init_fn_ -try: - from .flash_attn_triton import flash_attn_func -except: - pass -Tokenizer = Union[PreTrainedTokenizer, PreTrainedTokenizerFast] - -class MPTPreTrainedModel(PreTrainedModel): - config_class = MPTConfig - base_model_prefix = 'model' - _no_split_modules = ['MPTBlock'] - -class MPTModel(MPTPreTrainedModel): - - def __init__(self, config: MPTConfig): - config._validate_config() - super().__init__(config) - self.attn_impl = config.attn_config['attn_impl'] - self.prefix_lm = config.attn_config['prefix_lm'] - self.attn_uses_sequence_id = config.attn_config['attn_uses_sequence_id'] - self.alibi = config.attn_config['alibi'] - self.alibi_bias_max = config.attn_config['alibi_bias_max'] - if config.init_device == 'mixed': - if dist.get_local_rank() == 0: - config.init_device = 'cpu' - else: - config.init_device = 'meta' - if config.norm_type.lower() not in NORM_CLASS_REGISTRY.keys(): - norm_options = ' | '.join(NORM_CLASS_REGISTRY.keys()) - raise NotImplementedError(f'Requested norm type ({config.norm_type}) is not implemented within this repo (Options: {norm_options}).') - norm_class = NORM_CLASS_REGISTRY[config.norm_type.lower()] - self.embedding_fraction = config.embedding_fraction - self.wte = SharedEmbedding(config.vocab_size, config.d_model, device=config.init_device) - if not self.alibi: - self.wpe = torch.nn.Embedding(config.max_seq_len, config.d_model, device=config.init_device) - self.emb_drop = nn.Dropout(config.emb_pdrop) - self.blocks = nn.ModuleList([MPTBlock(device=config.init_device, **config.to_dict()) for _ in range(config.n_layers)]) - self.norm_f = norm_class(config.d_model, device=config.init_device) - if config.init_device != 'meta': - print(f'You are using config.init_device={config.init_device!r}, but you can also use config.init_device="meta" with Composer + FSDP for fast initialization.') - self.apply(self.param_init_fn) - self.is_causal = not self.prefix_lm - self._attn_bias_initialized = False - self.attn_bias = None - self.attn_bias_shape = attn_bias_shape(self.attn_impl, config.n_heads, config.max_seq_len, self.alibi, prefix_lm=self.prefix_lm, causal=self.is_causal, use_sequence_id=self.attn_uses_sequence_id) - if config.no_bias: - for module in self.modules(): - if hasattr(module, 'bias') and isinstance(module.bias, nn.Parameter): - if config.verbose: - warnings.warn(f'Removing bias ({module.bias}) from {module}.') - module.register_parameter('bias', None) - if config.verbose and config.verbose > 2: - print(self) - if 'verbose' not in self.config.init_config: - self.config.init_config['verbose'] = self.config.verbose - if self.config.init_config['verbose'] > 1: - init_fn_name = self.config.init_config['name'] - warnings.warn(f'Using {init_fn_name} initialization.') - self.gradient_checkpointing = False - - def get_input_embeddings(self): - return self.wte - - def set_input_embeddings(self, value): - self.wte = value - - @torch.no_grad() - def _attn_bias(self, device, dtype, attention_mask: Optional[torch.ByteTensor]=None, prefix_mask: Optional[torch.ByteTensor]=None, sequence_id: Optional[torch.LongTensor]=None): - if not self._attn_bias_initialized: - if self.attn_bias_shape: - self.attn_bias = torch.zeros(self.attn_bias_shape, device=device, dtype=dtype) - self.attn_bias = build_attn_bias(self.attn_impl, self.attn_bias, self.config.n_heads, self.config.max_seq_len, causal=self.is_causal, alibi=self.alibi, alibi_bias_max=self.alibi_bias_max) - self._attn_bias_initialized = True - if self.attn_impl == 'flash': - return (self.attn_bias, attention_mask) - if self.attn_bias is not None: - self.attn_bias = self.attn_bias.to(dtype=dtype, device=device) - attn_bias = self.attn_bias - if self.prefix_lm: - assert isinstance(attn_bias, torch.Tensor) - assert isinstance(prefix_mask, torch.Tensor) - attn_bias = self._apply_prefix_mask(attn_bias, prefix_mask) - if self.attn_uses_sequence_id and sequence_id is not None: - assert isinstance(attn_bias, torch.Tensor) - attn_bias = self._apply_sequence_id(attn_bias, sequence_id) - if attention_mask is not None: - s_k = attention_mask.shape[-1] - if attn_bias is None: - attn_bias = torch.zeros((1, 1, 1, s_k), device=device, dtype=dtype) - else: - _s_k = max(0, attn_bias.size(-1) - s_k) - attn_bias = attn_bias[:, :, :, _s_k:] - if prefix_mask is not None and attention_mask.shape != prefix_mask.shape: - raise ValueError(f'attention_mask shape={attention_mask.shape} ' + f'and prefix_mask shape={prefix_mask.shape} are not equal.') - min_val = torch.finfo(attn_bias.dtype).min - attn_bias = attn_bias.masked_fill(~attention_mask.view(-1, 1, 1, s_k), min_val) - return (attn_bias, None) - - def _apply_prefix_mask(self, attn_bias: torch.Tensor, prefix_mask: torch.Tensor): - (s_k, s_q) = attn_bias.shape[-2:] - if s_k != self.config.max_seq_len or s_q != self.config.max_seq_len: - raise ValueError('attn_bias does not match the expected shape. ' + f'The last two dimensions should both be {self.config.max_length} ' + f'but are {s_k} and {s_q}.') - seq_len = prefix_mask.shape[-1] - if seq_len > self.config.max_seq_len: - raise ValueError(f'prefix_mask sequence length cannot exceed max_seq_len={self.config.max_seq_len}') - attn_bias = attn_bias[..., :seq_len, :seq_len] - causal = torch.tril(torch.ones((seq_len, seq_len), dtype=torch.bool, device=prefix_mask.device)).view(1, 1, seq_len, seq_len) - prefix = prefix_mask.view(-1, 1, 1, seq_len) - cannot_attend = ~torch.logical_or(causal, prefix.bool()) - min_val = torch.finfo(attn_bias.dtype).min - attn_bias = attn_bias.masked_fill(cannot_attend, min_val) - return attn_bias - - def _apply_sequence_id(self, attn_bias: torch.Tensor, sequence_id: torch.LongTensor): - seq_len = sequence_id.shape[-1] - if seq_len > self.config.max_seq_len: - raise ValueError(f'sequence_id sequence length cannot exceed max_seq_len={self.config.max_seq_len}') - attn_bias = attn_bias[..., :seq_len, :seq_len] - cannot_attend = torch.logical_not(torch.eq(sequence_id.view(-1, seq_len, 1), sequence_id.view(-1, 1, seq_len))).unsqueeze(1) - min_val = torch.finfo(attn_bias.dtype).min - attn_bias = attn_bias.masked_fill(cannot_attend, min_val) - return attn_bias - - def forward(self, input_ids: torch.LongTensor, past_key_values: Optional[List[Tuple[torch.FloatTensor]]]=None, attention_mask: Optional[torch.ByteTensor]=None, prefix_mask: Optional[torch.ByteTensor]=None, sequence_id: Optional[torch.LongTensor]=None, return_dict: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, use_cache: Optional[bool]=None, inputs_embeds: Optional[torch.Tensor]=None): - return_dict = return_dict if return_dict is not None else self.config.return_dict - use_cache = use_cache if use_cache is not None else self.config.use_cache - if attention_mask is not None: - attention_mask = attention_mask.bool() - if prefix_mask is not None: - prefix_mask = prefix_mask.bool() - if not return_dict: - raise NotImplementedError('return_dict False is not implemented yet for MPT') - if output_attentions: - if self.attn_impl != 'torch': - raise NotImplementedError('output_attentions is not implemented for MPT when using attn_impl `flash` or `triton`.') - if attention_mask is not None and attention_mask[:, 0].sum() != attention_mask.shape[0] and self.training: - raise NotImplementedError('MPT does not support training with left padding.') - if self.prefix_lm and prefix_mask is None: - raise ValueError('prefix_mask is a required argument when MPT is configured with prefix_lm=True.') - if self.training: - if self.attn_uses_sequence_id and sequence_id is None: - raise ValueError('sequence_id is a required argument when MPT is configured with attn_uses_sequence_id=True ' + 'and the model is in train mode.') - elif self.attn_uses_sequence_id is False and sequence_id is not None: - warnings.warn('MPT received non-None input for `sequence_id` but is configured with attn_uses_sequence_id=False. ' + 'This input will be ignored. If you want the model to use `sequence_id`, set attn_uses_sequence_id to True.') - if input_ids is not None: - S = input_ids.size(1) - assert S <= self.config.max_seq_len, f'Cannot forward input with seq_len={S}, this model only supports seq_len<={self.config.max_seq_len}' - tok_emb = self.wte(input_ids) - else: - assert inputs_embeds is not None - assert self.alibi, 'inputs_embeds is not implemented for MPT unless for alibi.' - S = inputs_embeds.size(1) - tok_emb = inputs_embeds - if self.alibi: - x = tok_emb - else: - past_position = 0 - if past_key_values is not None: - if len(past_key_values) != self.config.n_layers: - raise ValueError(f'past_key_values must provide a past_key_value for each attention ' + f'layer in the network (len(past_key_values)={len(past_key_values)!r}; self.config.n_layers={self.config.n_layers!r}).') - past_position = past_key_values[0][0].size(1) - if self.attn_impl == 'torch': - past_position = past_key_values[0][0].size(3) - if S + past_position > self.config.max_seq_len: - raise ValueError(f'Cannot forward input with past sequence length {past_position} and current sequence length {S + 1}, this model only supports total sequence length <= {self.config.max_seq_len}.') - pos = torch.arange(past_position, S + past_position, dtype=torch.long, device=input_ids.device).unsqueeze(0) - if attention_mask is not None: - pos = torch.clamp(pos - torch.cumsum((~attention_mask).to(torch.int32), dim=1)[:, past_position:], min=0) - pos_emb = self.wpe(pos) - x = tok_emb + pos_emb - if self.embedding_fraction == 1: - x = self.emb_drop(x) - else: - x_shrunk = x * self.embedding_fraction + x.detach() * (1 - self.embedding_fraction) - assert isinstance(self.emb_drop, nn.Module) - x = self.emb_drop(x_shrunk) - (attn_bias, attention_mask) = self._attn_bias(device=x.device, dtype=torch.float32, attention_mask=attention_mask, prefix_mask=prefix_mask, sequence_id=sequence_id) - if use_cache and past_key_values is None: - past_key_values = [() for _ in range(self.config.n_layers)] - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for (b_idx, block) in enumerate(self.blocks): - if output_hidden_states: - assert all_hidden_states is not None - all_hidden_states = all_hidden_states + (x,) - past_key_value = past_key_values[b_idx] if past_key_values is not None else None - if self.gradient_checkpointing and self.training: - (x, attn_weights, past_key_value) = torch.utils.checkpoint.checkpoint(block, x, past_key_value, attn_bias, attention_mask, self.is_causal) - else: - (x, attn_weights, past_key_value) = block(x, past_key_value=past_key_value, attn_bias=attn_bias, attention_mask=attention_mask, is_causal=self.is_causal) - if past_key_values is not None: - past_key_values[b_idx] = past_key_value - if output_attentions: - assert all_self_attns is not None - all_self_attns = all_self_attns + (attn_weights,) - x = self.norm_f(x) - if output_hidden_states: - assert all_hidden_states is not None - all_hidden_states = all_hidden_states + (x,) - return BaseModelOutputWithPast(last_hidden_state=x, past_key_values=past_key_values, hidden_states=all_hidden_states, attentions=all_self_attns) - - def param_init_fn(self, module): - init_fn_name = self.config.init_config['name'] - MODEL_INIT_REGISTRY[init_fn_name](module=module, n_layers=self.config.n_layers, d_model=self.config.d_model, **self.config.init_config) - - def fsdp_wrap_fn(self, module): - return isinstance(module, MPTBlock) - - def activation_checkpointing_fn(self, module): - return isinstance(module, MPTBlock) - -class MPTForCausalLM(MPTPreTrainedModel): - - def __init__(self, config: MPTConfig): - super().__init__(config) - if not config.tie_word_embeddings: - raise ValueError('MPTForCausalLM only supports tied word embeddings') - print(f'Instantiating an MPTForCausalLM model from {__file__}') - self.transformer = MPTModel(config) - for child in self.transformer.children(): - if isinstance(child, torch.nn.ModuleList): - continue - if isinstance(child, torch.nn.Module): - child._fsdp_wrap = True - self.logit_scale = None - if config.logit_scale is not None: - logit_scale = config.logit_scale - if isinstance(logit_scale, str): - if logit_scale == 'inv_sqrt_d_model': - logit_scale = 1 / math.sqrt(config.d_model) - else: - raise ValueError(f"logit_scale={logit_scale!r} is not recognized as an option; use numeric value or 'inv_sqrt_d_model'.") - self.logit_scale = logit_scale - - def get_input_embeddings(self): - return self.transformer.wte - - def set_input_embeddings(self, value): - self.transformer.wte = value - - def get_output_embeddings(self): - return self.transformer.wte - - def set_output_embeddings(self, new_embeddings): - self.transformer.wte = new_embeddings - - def set_decoder(self, decoder): - self.transformer = decoder - - def get_decoder(self): - return self.transformer - - def forward(self, input_ids: torch.LongTensor, past_key_values: Optional[List[Tuple[torch.FloatTensor]]]=None, attention_mask: Optional[torch.ByteTensor]=None, prefix_mask: Optional[torch.ByteTensor]=None, sequence_id: Optional[torch.LongTensor]=None, labels: Optional[torch.LongTensor]=None, return_dict: Optional[bool]=None, output_attentions: Optional[bool]=None, output_hidden_states: Optional[bool]=None, use_cache: Optional[bool]=None, inputs_embeds: Optional[torch.FloatTensor]=None): - return_dict = return_dict if return_dict is not None else self.config.return_dict - use_cache = use_cache if use_cache is not None else self.config.use_cache - if inputs_embeds is not None: - raise NotImplementedError('inputs_embeds has to be None (for hf/peft support).') - outputs = self.transformer(input_ids=input_ids, past_key_values=past_key_values, attention_mask=attention_mask, prefix_mask=prefix_mask, sequence_id=sequence_id, return_dict=return_dict, output_attentions=output_attentions, output_hidden_states=output_hidden_states, use_cache=use_cache) - logits = self.transformer.wte(outputs.last_hidden_state.to(self.transformer.wte.weight.device), True) - if self.logit_scale is not None: - if self.logit_scale == 0: - warnings.warn(f'Multiplying logits by self.logit_scale={self.logit_scale!r}. This will produce uniform (uninformative) outputs.') - logits *= self.logit_scale - loss = None - if labels is not None: - labels = torch.roll(labels, shifts=-1) - labels[:, -1] = -100 - loss = F.cross_entropy(logits.view(-1, logits.size(-1)), labels.to(logits.device).view(-1)) - return CausalLMOutputWithPast(loss=loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions) - - def param_init_fn(self, module): - init_fn_name = self.config.init_config['name'] - MODEL_INIT_REGISTRY[init_fn_name](module=module, n_layers=self.config.n_layers, d_model=self.config.d_model, **self.config.init_config) - - def fsdp_wrap_fn(self, module): - return isinstance(module, MPTBlock) - - def activation_checkpointing_fn(self, module): - return isinstance(module, MPTBlock) - - def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs): - if inputs_embeds is not None: - raise NotImplementedError('inputs_embeds is not implemented for MPT yet') - attention_mask = kwargs['attention_mask'].bool() - if attention_mask[:, -1].sum() != attention_mask.shape[0]: - raise NotImplementedError('MPT does not support generation with right padding.') - if self.transformer.attn_uses_sequence_id and self.training: - sequence_id = torch.zeros_like(input_ids[:1]) - else: - sequence_id = None - if past_key_values is not None: - input_ids = input_ids[:, -1].unsqueeze(-1) - if self.transformer.prefix_lm: - prefix_mask = torch.ones_like(attention_mask) - if kwargs.get('use_cache') == False: - raise NotImplementedError('MPT with prefix_lm=True does not support use_cache=False.') - else: - prefix_mask = None - return {'input_ids': input_ids, 'attention_mask': attention_mask, 'prefix_mask': prefix_mask, 'sequence_id': sequence_id, 'past_key_values': past_key_values, 'use_cache': kwargs.get('use_cache', True)} - - @staticmethod - def _reorder_cache(past_key_values, beam_idx): - """Used by HuggingFace generate when using beam search with kv-caching. - - See https://github.com/huggingface/transformers/blob/3ec7a47664ebe40c40f4b722f6bb1cd30c3821ec/src/transformers/models/gpt2/modeling_gpt2.py#L1122-L1133 - for an example in transformers. - """ - reordered_past = [] - for layer_past in past_key_values: - reordered_past += [tuple((past_state.index_select(0, beam_idx) for past_state in layer_past))] - return reordered_past \ No newline at end of file diff --git a/llava/model/language_model/mpt/norm.py b/llava/model/language_model/mpt/norm.py deleted file mode 100644 index 067b6140f..000000000 --- a/llava/model/language_model/mpt/norm.py +++ /dev/null @@ -1,56 +0,0 @@ -import torch - -def _cast_if_autocast_enabled(tensor): - if torch.is_autocast_enabled(): - if tensor.device.type == 'cuda': - dtype = torch.get_autocast_gpu_dtype() - elif tensor.device.type == 'cpu': - dtype = torch.get_autocast_cpu_dtype() - else: - raise NotImplementedError() - return tensor.to(dtype=dtype) - return tensor - -class LPLayerNorm(torch.nn.LayerNorm): - - def __init__(self, normalized_shape, eps=1e-05, elementwise_affine=True, device=None, dtype=None): - super().__init__(normalized_shape=normalized_shape, eps=eps, elementwise_affine=elementwise_affine, device=device, dtype=dtype) - - def forward(self, x): - module_device = x.device - downcast_x = _cast_if_autocast_enabled(x) - downcast_weight = _cast_if_autocast_enabled(self.weight) if self.weight is not None else self.weight - downcast_bias = _cast_if_autocast_enabled(self.bias) if self.bias is not None else self.bias - with torch.autocast(enabled=False, device_type=module_device.type): - return torch.nn.functional.layer_norm(downcast_x, self.normalized_shape, downcast_weight, downcast_bias, self.eps) - -def rms_norm(x, weight=None, eps=1e-05): - output = x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + eps) - if weight is not None: - return output * weight - return output - -class RMSNorm(torch.nn.Module): - - def __init__(self, normalized_shape, eps=1e-05, weight=True, dtype=None, device=None): - super().__init__() - self.eps = eps - if weight: - self.weight = torch.nn.Parameter(torch.ones(normalized_shape, dtype=dtype, device=device)) - else: - self.register_parameter('weight', None) - - def forward(self, x): - return rms_norm(x.float(), self.weight, self.eps).to(dtype=x.dtype) - -class LPRMSNorm(RMSNorm): - - def __init__(self, normalized_shape, eps=1e-05, weight=True, dtype=None, device=None): - super().__init__(normalized_shape=normalized_shape, eps=eps, weight=weight, dtype=dtype, device=device) - - def forward(self, x): - downcast_x = _cast_if_autocast_enabled(x) - downcast_weight = _cast_if_autocast_enabled(self.weight) if self.weight is not None else self.weight - with torch.autocast(enabled=False, device_type=x.device.type): - return rms_norm(downcast_x, downcast_weight, self.eps).to(dtype=x.dtype) -NORM_CLASS_REGISTRY = {'layernorm': torch.nn.LayerNorm, 'low_precision_layernorm': LPLayerNorm, 'rmsnorm': RMSNorm, 'low_precision_rmsnorm': LPRMSNorm} \ No newline at end of file diff --git a/llava/model/language_model/mpt/param_init_fns.py b/llava/model/language_model/mpt/param_init_fns.py deleted file mode 100644 index 418b83ca2..000000000 --- a/llava/model/language_model/mpt/param_init_fns.py +++ /dev/null @@ -1,181 +0,0 @@ -import math -import warnings -from collections.abc import Sequence -from functools import partial -from typing import Optional, Tuple, Union -import torch -from torch import nn -from .norm import NORM_CLASS_REGISTRY - -def torch_default_param_init_fn_(module: nn.Module, verbose: int=0, **kwargs): - del kwargs - if verbose > 1: - warnings.warn(f"Initializing network using module's reset_parameters attribute") - if hasattr(module, 'reset_parameters'): - module.reset_parameters() - -def fused_init_helper_(module: nn.Module, init_fn_): - _fused = getattr(module, '_fused', None) - if _fused is None: - raise RuntimeError(f'Internal logic error') - (dim, splits) = _fused - splits = (0, *splits, module.weight.size(dim)) - for (s, e) in zip(splits[:-1], splits[1:]): - slice_indices = [slice(None)] * module.weight.ndim - slice_indices[dim] = slice(s, e) - init_fn_(module.weight[slice_indices]) - -def generic_param_init_fn_(module: nn.Module, init_fn_, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, verbose: int=0, **kwargs): - del kwargs - if verbose > 1: - warnings.warn(f'If model has bias parameters they are initialized to 0.') - init_div_is_residual = init_div_is_residual - if init_div_is_residual is False: - div_is_residual = 1.0 - elif init_div_is_residual is True: - div_is_residual = math.sqrt(2 * n_layers) - elif isinstance(init_div_is_residual, float) or isinstance(init_div_is_residual, int): - div_is_residual = init_div_is_residual - elif isinstance(init_div_is_residual, str) and init_div_is_residual.isnumeric(): - div_is_residual = float(init_div_is_residual) - else: - div_is_residual = 1.0 - raise ValueError(f'Expected init_div_is_residual to be boolean or numeric, got {init_div_is_residual}') - if init_div_is_residual is not False: - if verbose > 1: - warnings.warn(f'Initializing _is_residual layers then dividing them by {div_is_residual:.3f}. ' + f'Set `init_div_is_residual: false` in init config to disable this.') - if isinstance(module, nn.Linear): - if hasattr(module, '_fused'): - fused_init_helper_(module, init_fn_) - else: - init_fn_(module.weight) - if module.bias is not None: - torch.nn.init.zeros_(module.bias) - if init_div_is_residual is not False and getattr(module, '_is_residual', False): - with torch.no_grad(): - module.weight.div_(div_is_residual) - elif isinstance(module, nn.Embedding): - if emb_init_std is not None: - std = emb_init_std - if std == 0: - warnings.warn(f'Embedding layer initialized to 0.') - emb_init_fn_ = partial(torch.nn.init.normal_, mean=0.0, std=std) - if verbose > 1: - warnings.warn(f'Embedding layer initialized using normal distribution with mean=0 and std={std!r}.') - elif emb_init_uniform_lim is not None: - lim = emb_init_uniform_lim - if isinstance(lim, Sequence): - if len(lim) > 2: - raise ValueError(f'Uniform init requires a min and a max limit. User input: {lim}.') - if lim[0] == lim[1]: - warnings.warn(f'Embedding layer initialized to {lim[0]}.') - else: - if lim == 0: - warnings.warn(f'Embedding layer initialized to 0.') - lim = [-lim, lim] - (a, b) = lim - emb_init_fn_ = partial(torch.nn.init.uniform_, a=a, b=b) - if verbose > 1: - warnings.warn(f'Embedding layer initialized using uniform distribution in range {lim}.') - else: - emb_init_fn_ = init_fn_ - emb_init_fn_(module.weight) - elif isinstance(module, tuple(set(NORM_CLASS_REGISTRY.values()))): - if verbose > 1: - warnings.warn(f'Norm weights are set to 1. If norm layer has a bias it is initialized to 0.') - if hasattr(module, 'weight') and module.weight is not None: - torch.nn.init.ones_(module.weight) - if hasattr(module, 'bias') and module.bias is not None: - torch.nn.init.zeros_(module.bias) - elif isinstance(module, nn.MultiheadAttention): - if module._qkv_same_embed_dim: - assert module.in_proj_weight is not None - assert module.q_proj_weight is None and module.k_proj_weight is None and (module.v_proj_weight is None) - assert d_model is not None - _d = d_model - splits = (0, _d, 2 * _d, 3 * _d) - for (s, e) in zip(splits[:-1], splits[1:]): - init_fn_(module.in_proj_weight[s:e]) - else: - assert module.q_proj_weight is not None and module.k_proj_weight is not None and (module.v_proj_weight is not None) - assert module.in_proj_weight is None - init_fn_(module.q_proj_weight) - init_fn_(module.k_proj_weight) - init_fn_(module.v_proj_weight) - if module.in_proj_bias is not None: - torch.nn.init.zeros_(module.in_proj_bias) - if module.bias_k is not None: - torch.nn.init.zeros_(module.bias_k) - if module.bias_v is not None: - torch.nn.init.zeros_(module.bias_v) - init_fn_(module.out_proj.weight) - if init_div_is_residual is not False and getattr(module.out_proj, '_is_residual', False): - with torch.no_grad(): - module.out_proj.weight.div_(div_is_residual) - if module.out_proj.bias is not None: - torch.nn.init.zeros_(module.out_proj.bias) - else: - for _ in module.parameters(recurse=False): - raise NotImplementedError(f'{module.__class__.__name__} parameters are not initialized by param_init_fn.') - -def _normal_init_(std, mean=0.0): - return partial(torch.nn.init.normal_, mean=mean, std=std) - -def _normal_param_init_fn_(module: nn.Module, std: float, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, verbose: int=0, **kwargs): - del kwargs - init_fn_ = _normal_init_(std=std) - if verbose > 1: - warnings.warn(f'Using torch.nn.init.normal_ init fn mean=0.0, std={std}') - generic_param_init_fn_(module=module, init_fn_=init_fn_, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def baseline_param_init_fn_(module: nn.Module, init_std: float, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, verbose: int=0, **kwargs): - del kwargs - if init_std is None: - raise ValueError("You must set model.init_config['init_std'] to a float value to use the default initialization scheme.") - _normal_param_init_fn_(module=module, std=init_std, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def small_param_init_fn_(module: nn.Module, n_layers: int, d_model: int, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, verbose: int=0, **kwargs): - del kwargs - std = math.sqrt(2 / (5 * d_model)) - _normal_param_init_fn_(module=module, std=std, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def neox_param_init_fn_(module: nn.Module, n_layers: int, d_model: int, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, verbose: int=0, **kwargs): - """From section 2.3.1 of GPT-NeoX-20B: - - An Open-Source AutoregressiveLanguage Model — Black et. al. (2022) - see https://github.com/EleutherAI/gpt-neox/blob/9610391ab319403cef079b438edd016a2443af54/megatron/model/init_functions.py#L151 - and https://github.com/EleutherAI/gpt-neox/blob/main/megatron/model/transformer.py - """ - del kwargs - residual_div = n_layers / math.sqrt(10) - if verbose > 1: - warnings.warn(f'setting init_div_is_residual to {residual_div}') - small_param_init_fn_(module=module, d_model=d_model, n_layers=n_layers, init_div_is_residual=residual_div, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def kaiming_uniform_param_init_fn_(module: nn.Module, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, init_gain: float=0, fan_mode: str='fan_in', init_nonlinearity: str='leaky_relu', verbose: int=0, **kwargs): - del kwargs - if verbose > 1: - warnings.warn(f'Using nn.init.kaiming_uniform_ init fn with parameters: ' + f'a={init_gain}, mode={fan_mode}, nonlinearity={init_nonlinearity}') - kaiming_uniform_ = partial(nn.init.kaiming_uniform_, a=init_gain, mode=fan_mode, nonlinearity=init_nonlinearity) - generic_param_init_fn_(module=module, init_fn_=kaiming_uniform_, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def kaiming_normal_param_init_fn_(module: nn.Module, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, init_gain: float=0, fan_mode: str='fan_in', init_nonlinearity: str='leaky_relu', verbose: int=0, **kwargs): - del kwargs - if verbose > 1: - warnings.warn(f'Using nn.init.kaiming_normal_ init fn with parameters: ' + f'a={init_gain}, mode={fan_mode}, nonlinearity={init_nonlinearity}') - kaiming_normal_ = partial(torch.nn.init.kaiming_normal_, a=init_gain, mode=fan_mode, nonlinearity=init_nonlinearity) - generic_param_init_fn_(module=module, init_fn_=kaiming_normal_, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def xavier_uniform_param_init_fn_(module: nn.Module, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, init_gain: float=0, verbose: int=0, **kwargs): - del kwargs - xavier_uniform_ = partial(torch.nn.init.xavier_uniform_, gain=init_gain) - if verbose > 1: - warnings.warn(f'Using torch.nn.init.xavier_uniform_ init fn with parameters: ' + f'gain={init_gain}') - generic_param_init_fn_(module=module, init_fn_=xavier_uniform_, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) - -def xavier_normal_param_init_fn_(module: nn.Module, n_layers: int, d_model: Optional[int]=None, init_div_is_residual: Union[int, float, str, bool]=True, emb_init_std: Optional[float]=None, emb_init_uniform_lim: Optional[Union[Tuple[float, float], float]]=None, init_gain: float=0, verbose: int=0, **kwargs): - xavier_normal_ = partial(torch.nn.init.xavier_normal_, gain=init_gain) - if verbose > 1: - warnings.warn(f'Using torch.nn.init.xavier_normal_ init fn with parameters: ' + f'gain={init_gain}') - generic_param_init_fn_(module=module, init_fn_=xavier_normal_, d_model=d_model, n_layers=n_layers, init_div_is_residual=init_div_is_residual, emb_init_std=emb_init_std, emb_init_uniform_lim=emb_init_uniform_lim, verbose=verbose) -MODEL_INIT_REGISTRY = {'default_': torch_default_param_init_fn_, 'baseline_': baseline_param_init_fn_, 'kaiming_uniform_': kaiming_uniform_param_init_fn_, 'kaiming_normal_': kaiming_normal_param_init_fn_, 'neox_init_': neox_param_init_fn_, 'small_init_': small_param_init_fn_, 'xavier_uniform_': xavier_uniform_param_init_fn_, 'xavier_normal_': xavier_normal_param_init_fn_} \ No newline at end of file diff --git a/llava/model/llava_arch.py b/llava/model/llava_arch.py index d2f396f39..5a5e62dfc 100644 --- a/llava/model/llava_arch.py +++ b/llava/model/llava_arch.py @@ -23,6 +23,8 @@ from llava.constants import IGNORE_INDEX, IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_PATCH_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN +from llava.mm_utils import get_anyres_image_grid_shape + class LlavaMetaModel: @@ -33,6 +35,11 @@ def __init__(self, config): self.vision_tower = build_vision_tower(config, delay_load=True) self.mm_projector = build_vision_projector(config) + if 'unpad' in getattr(config, 'mm_patch_merge_type', ''): + self.image_newline = nn.Parameter( + torch.empty(config.hidden_size, dtype=self.dtype) + ) + def get_vision_tower(self): vision_tower = getattr(self, 'vision_tower', None) if type(vision_tower) is list: @@ -44,6 +51,7 @@ def initialize_vision_modules(self, model_args, fsdp=None): mm_vision_select_layer = model_args.mm_vision_select_layer mm_vision_select_feature = model_args.mm_vision_select_feature pretrain_mm_mlp_adapter = model_args.pretrain_mm_mlp_adapter + mm_patch_merge_type = model_args.mm_patch_merge_type self.config.mm_vision_tower = vision_tower @@ -66,9 +74,16 @@ def initialize_vision_modules(self, model_args, fsdp=None): self.config.mm_hidden_size = vision_tower.hidden_size self.config.mm_vision_select_layer = mm_vision_select_layer self.config.mm_vision_select_feature = mm_vision_select_feature + self.config.mm_patch_merge_type = mm_patch_merge_type if getattr(self, 'mm_projector', None) is None: self.mm_projector = build_vision_projector(self.config) + + if 'unpad' in mm_patch_merge_type: + embed_std = 1 / torch.sqrt(torch.tensor(self.config.hidden_size, dtype=self.dtype)) + self.image_newline = nn.Parameter( + torch.randn(self.config.hidden_size, dtype=self.dtype) * embed_std + ) else: # In case it is frozen by LoRA for p in self.mm_projector.parameters(): @@ -82,6 +97,37 @@ def get_w(weights, keyword): self.mm_projector.load_state_dict(get_w(mm_projector_weights, 'mm_projector')) +def unpad_image(tensor, original_size): + """ + Unpads a PyTorch tensor of a padded and resized image. + + Args: + tensor (torch.Tensor): The image tensor, assumed to be in CxHxW format. + original_size (tuple): The original size of the image (height, width). + + Returns: + torch.Tensor: The unpadded image tensor. + """ + original_width, original_height = original_size + current_height, current_width = tensor.shape[1:] + + original_aspect_ratio = original_width / original_height + current_aspect_ratio = current_width / current_height + + if original_aspect_ratio > current_aspect_ratio: + scale_factor = current_width / original_width + new_height = int(original_height * scale_factor) + padding = (current_height - new_height) // 2 + unpadded_tensor = tensor[:, padding:current_height - padding, :] + else: + scale_factor = current_height / original_height + new_width = int(original_width * scale_factor) + padding = (current_width - new_width) // 2 + unpadded_tensor = tensor[:, :, padding:current_width - padding] + + return unpadded_tensor + + class LlavaMetaForCausalLM(ABC): @abstractmethod @@ -97,28 +143,63 @@ def encode_images(self, images): return image_features def prepare_inputs_labels_for_multimodal( - self, input_ids, position_ids, attention_mask, past_key_values, labels, images + self, input_ids, position_ids, attention_mask, past_key_values, labels, + images, image_sizes=None ): vision_tower = self.get_vision_tower() if vision_tower is None or images is None or input_ids.shape[1] == 1: - if past_key_values is not None and vision_tower is not None and images is not None and input_ids.shape[1] == 1: - target_shape = past_key_values[-1][-1].shape[-2] + 1 - attention_mask = torch.cat((attention_mask, torch.ones( - (attention_mask.shape[0], target_shape - attention_mask.shape[1]), - dtype=attention_mask.dtype, - device=attention_mask.device - )), dim=1) - position_ids = torch.sum(attention_mask, dim=1).unsqueeze(-1) - 1 return input_ids, position_ids, attention_mask, past_key_values, None, labels if type(images) is list or images.ndim == 5: + if type(images) is list: + images = [x.unsqueeze(0) if x.ndim == 3 else x for x in images] concat_images = torch.cat([image for image in images], dim=0) image_features = self.encode_images(concat_images) split_sizes = [image.shape[0] for image in images] image_features = torch.split(image_features, split_sizes, dim=0) - image_features = [x.flatten(0, 1).to(self.device) for x in image_features] + mm_patch_merge_type = getattr(self.config, 'mm_patch_merge_type', 'flat') + image_aspect_ratio = getattr(self.config, 'image_aspect_ratio', 'square') + if mm_patch_merge_type == 'flat': + image_features = [x.flatten(0, 1) for x in image_features] + elif mm_patch_merge_type.startswith('spatial'): + new_image_features = [] + for image_idx, image_feature in enumerate(image_features): + if image_feature.shape[0] > 1: + base_image_feature = image_feature[0] + image_feature = image_feature[1:] + height = width = self.get_vision_tower().num_patches_per_side + assert height * width == base_image_feature.shape[0] + if image_aspect_ratio == 'anyres': + num_patch_width, num_patch_height = get_anyres_image_grid_shape(image_sizes[image_idx], self.config.image_grid_pinpoints, self.get_vision_tower().config.image_size) + image_feature = image_feature.view(num_patch_height, num_patch_width, height, width, -1) + else: + raise NotImplementedError + if 'unpad' in mm_patch_merge_type: + image_feature = image_feature.permute(4, 0, 2, 1, 3).contiguous() + image_feature = image_feature.flatten(1, 2).flatten(2, 3) + image_feature = unpad_image(image_feature, image_sizes[image_idx]) + image_feature = torch.cat(( + image_feature, + self.model.image_newline[:, None, None].expand(*image_feature.shape[:-1], 1) + ), dim=-1) + image_feature = image_feature.flatten(1, 2).transpose(0, 1) + else: + image_feature = image_feature.permute(0, 2, 1, 3, 4).contiguous() + image_feature = image_feature.flatten(0, 3) + image_feature = torch.cat((base_image_feature, image_feature), dim=0) + else: + image_feature = image_feature[0] + if 'unpad' in mm_patch_merge_type: + image_feature = torch.cat(( + image_feature, + self.model.image_newline[None] + ), dim=0) + new_image_features.append(image_feature) + image_features = new_image_features + else: + raise ValueError(f"Unexpected mm_patch_merge_type: {self.config.mm_patch_merge_type}") else: - image_features = self.encode_images(images).to(self.device) + image_features = self.encode_images(images) # TODO: image start / end is not implemented here to support pretraining. if getattr(self.config, 'tune_mm_mlp_adapter', False) and getattr(self.config, 'mm_use_im_start_end', False): @@ -140,7 +221,8 @@ def prepare_inputs_labels_for_multimodal( if labels is None: labels = torch.full_like(input_ids, IGNORE_INDEX) - # remove the padding using attention_mask -- TODO: double check + # remove the padding using attention_mask -- FIXME + _input_ids = input_ids input_ids = [cur_input_ids[cur_attention_mask] for cur_input_ids, cur_attention_mask in zip(input_ids, attention_mask)] labels = [cur_labels[cur_attention_mask] for cur_labels, cur_attention_mask in zip(labels, attention_mask)] @@ -180,6 +262,8 @@ def prepare_inputs_labels_for_multimodal( cur_new_input_embeds.append(cur_image_features) cur_new_labels.append(torch.full((cur_image_features.shape[0],), IGNORE_INDEX, device=cur_labels.device, dtype=cur_labels.dtype)) + cur_new_input_embeds = [x.to(self.device) for x in cur_new_input_embeds] + cur_new_input_embeds = torch.cat(cur_new_input_embeds) cur_new_labels = torch.cat(cur_new_labels) diff --git a/llava/model/multimodal_encoder/builder.py b/llava/model/multimodal_encoder/builder.py index 2b13589d4..e89507c49 100644 --- a/llava/model/multimodal_encoder/builder.py +++ b/llava/model/multimodal_encoder/builder.py @@ -5,7 +5,7 @@ def build_vision_tower(vision_tower_cfg, **kwargs): vision_tower = getattr(vision_tower_cfg, 'mm_vision_tower', getattr(vision_tower_cfg, 'vision_tower', None)) is_absolute_path_exists = os.path.exists(vision_tower) - if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion"): + if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion") or "ShareGPT4V" in vision_tower: return CLIPVisionTower(vision_tower, args=vision_tower_cfg, **kwargs) raise ValueError(f'Unknown vision tower: {vision_tower}') diff --git a/llava/model/multimodal_encoder/clip_encoder.py b/llava/model/multimodal_encoder/clip_encoder.py index dbb9015b0..211781135 100644 --- a/llava/model/multimodal_encoder/clip_encoder.py +++ b/llava/model/multimodal_encoder/clip_encoder.py @@ -16,6 +16,8 @@ def __init__(self, vision_tower, args, delay_load=False): if not delay_load: self.load_model() + elif getattr(args, 'unfreeze_mm_vision_tower', False): + self.load_model() else: self.cfg_only = CLIPVisionConfig.from_pretrained(self.vision_tower_name) @@ -73,6 +75,10 @@ def config(self): def hidden_size(self): return self.config.hidden_size + @property + def num_patches_per_side(self): + return self.config.image_size // self.config.patch_size + @property def num_patches(self): return (self.config.image_size // self.config.patch_size) ** 2 diff --git a/llava/serve/gradio_web_server.py b/llava/serve/gradio_web_server.py index e506334bd..0a7145421 100644 --- a/llava/serve/gradio_web_server.py +++ b/llava/serve/gradio_web_server.py @@ -168,6 +168,15 @@ def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request: if "llava" in model_name.lower(): if 'llama-2' in model_name.lower(): template_name = "llava_llama_2" + elif "mistral" in model_name.lower() or "mixtral" in model_name.lower(): + if 'orca' in model_name.lower(): + template_name = "mistral_orca" + elif 'hermes' in model_name.lower(): + template_name = "chatml_direct" + else: + template_name = "mistral_instruct" + elif 'llava-v1.6-34b' in model_name.lower(): + template_name = "chatml_direct" elif "v1" in model_name.lower(): if 'mmtag' in model_name.lower(): template_name = "v1_mmtag" @@ -280,7 +289,7 @@ def http_bot(state, model_selector, temperature, top_p, max_new_tokens, request: title_markdown = (""" # 🌋 LLaVA: Large Language and Vision Assistant -[[Project Page](https://llava-vl.github.io)] [[Code](https://github.com/haotian-liu/LLaVA)] [[Model](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md)] | 📚 [[LLaVA](https://arxiv.org/abs/2304.08485)] [[LLaVA-v1.5](https://arxiv.org/abs/2310.03744)] +[[Project Page](https://llava-vl.github.io)] [[Code](https://github.com/haotian-liu/LLaVA)] [[Model](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md)] | 📚 [[LLaVA](https://arxiv.org/abs/2304.08485)] [[LLaVA-v1.5](https://arxiv.org/abs/2310.03744)] [[LLaVA-v1.6](https://llava-vl.github.io/blog/2024-01-30-llava-1-6/)] """) tos_markdown = (""" diff --git a/llava/serve/model_worker.py b/llava/serve/model_worker.py index a7bcd0829..dc897ef2d 100644 --- a/llava/serve/model_worker.py +++ b/llava/serve/model_worker.py @@ -133,6 +133,7 @@ def generate_stream(self, params): raise ValueError("Number of images does not match number of tokens in prompt") images = [load_image_from_base64(image) for image in images] + image_sizes = [image.size for image in images] images = process_images(images, image_processor, model.config) if type(images) is list: @@ -148,7 +149,8 @@ def generate_stream(self, params): num_image_tokens = prompt.count(replace_token) * model.get_vision_tower().num_patches else: images = None - image_args = {"images": images} + image_sizes = None + image_args = {"images": images, "image_sizes": image_sizes} else: images = None image_args = {} diff --git a/llava/serve/sglang_worker.py b/llava/serve/sglang_worker.py new file mode 100644 index 000000000..fb7a555fc --- /dev/null +++ b/llava/serve/sglang_worker.py @@ -0,0 +1,244 @@ +""" +A model worker executes the model. +""" +import argparse +import asyncio +from concurrent.futures import ThreadPoolExecutor +import json +import time +import threading +import uuid + +from fastapi import FastAPI, Request, BackgroundTasks +from fastapi.responses import StreamingResponse +import requests +import re +import uvicorn +from functools import partial + +from llava.constants import WORKER_HEART_BEAT_INTERVAL +from llava.utils import (build_logger, server_error_msg, + pretty_print_semaphore) +from llava.mm_utils import process_images, load_image_from_base64, tokenizer_image_token, expand2square +from llava.constants import DEFAULT_IMAGE_TOKEN + +import sglang as sgl +from sglang.backend.runtime_endpoint import RuntimeEndpoint + + +GB = 1 << 30 + +worker_id = str(uuid.uuid4())[:6] +logger = build_logger("model_worker", f"model_worker_{worker_id}.log") +global_counter = 0 + +model_semaphore = None + + +def heart_beat_worker(controller): + while True: + time.sleep(WORKER_HEART_BEAT_INTERVAL) + controller.send_heart_beat() + + +@sgl.function +def pipeline(s, prompt, max_tokens): + for p in prompt: + if type(p) is str: + s += p + else: + s += sgl.image(p) + s += sgl.gen("response", max_tokens=max_tokens) + + +class ModelWorker: + def __init__(self, controller_addr, worker_addr, sgl_endpoint, + worker_id, no_register, model_name): + self.controller_addr = controller_addr + self.worker_addr = worker_addr + self.worker_id = worker_id + + # Select backend + backend = RuntimeEndpoint(sgl_endpoint) + sgl.set_default_backend(backend) + model_path = backend.model_info["model_path"] + + if model_path.endswith("/"): + model_path = model_path[:-1] + if model_name is None: + model_paths = model_path.split("/") + if model_paths[-1].startswith('checkpoint-'): + self.model_name = model_paths[-2] + "_" + model_paths[-1] + else: + self.model_name = model_paths[-1] + else: + self.model_name = model_name + + logger.info(f"Loading the SGLANG model {self.model_name} on worker {worker_id} ...") + + if not no_register: + self.register_to_controller() + self.heart_beat_thread = threading.Thread( + target=heart_beat_worker, args=(self,)) + self.heart_beat_thread.start() + + def register_to_controller(self): + logger.info("Register to controller") + + url = self.controller_addr + "/register_worker" + data = { + "worker_name": self.worker_addr, + "check_heart_beat": True, + "worker_status": self.get_status() + } + r = requests.post(url, json=data) + assert r.status_code == 200 + + def send_heart_beat(self): + logger.info(f"Send heart beat. Models: {[self.model_name]}. " + f"Semaphore: {pretty_print_semaphore(model_semaphore)}. " + f"global_counter: {global_counter}") + + url = self.controller_addr + "/receive_heart_beat" + + while True: + try: + ret = requests.post(url, json={ + "worker_name": self.worker_addr, + "queue_length": self.get_queue_length()}, timeout=5) + exist = ret.json()["exist"] + break + except requests.exceptions.RequestException as e: + logger.error(f"heart beat error: {e}") + time.sleep(5) + + if not exist: + self.register_to_controller() + + def get_queue_length(self): + if model_semaphore is None: + return 0 + else: + return args.limit_model_concurrency - model_semaphore._value + (len( + model_semaphore._waiters) if model_semaphore._waiters is not None else 0) + + def get_status(self): + return { + "model_names": [self.model_name], + "speed": 1, + "queue_length": self.get_queue_length(), + } + + async def generate_stream(self, params): + ori_prompt = prompt = params["prompt"] + images = params.get("images", None) + if images is not None and len(images) > 0: + if len(images) > 0: + if len(images) != prompt.count(DEFAULT_IMAGE_TOKEN): + raise ValueError("Number of images does not match number of tokens in prompt") + + images = [load_image_from_base64(image) for image in images] + + # FIXME: for image-start/end token + # replace_token = DEFAULT_IMAGE_TOKEN + # if getattr(self.model.config, 'mm_use_im_start_end', False): + # replace_token = DEFAULT_IM_START_TOKEN + replace_token + DEFAULT_IM_END_TOKEN + # prompt = prompt.replace(DEFAULT_IMAGE_TOKEN, replace_token) + prompt = prompt.replace(' ' + DEFAULT_IMAGE_TOKEN + '\n', DEFAULT_IMAGE_TOKEN) + prompt_split = prompt.split(DEFAULT_IMAGE_TOKEN) + prompt = [] + for i in range(len(prompt_split)): + prompt.append(prompt_split[i]) + if i < len(images): + prompt.append(images[i]) + else: + prompt = [prompt] + + temperature = float(params.get("temperature", 1.0)) + top_p = float(params.get("top_p", 1.0)) + # max_context_length = getattr(model.config, 'max_position_embeddings', 2048) + max_new_tokens = min(int(params.get("max_new_tokens", 256)), 1024) + stop_str = params.get("stop", None) + stop_str = [stop_str] if stop_str is not None else None + + print({'prompt': prompt, 'max_new_tokens': max_new_tokens, 'temperature': temperature, 'top_p': top_p}) + state = pipeline.run(prompt, max_new_tokens, temperature=temperature, top_p=top_p, stream=True) + + generated_text = ori_prompt + async for text_outputs in state.text_async_iter(var_name="response"): + generated_text += text_outputs + yield json.dumps({"text": generated_text, "error_code": 0}).encode() + b"\0" + + async def generate_stream_gate(self, params): + try: + async for x in self.generate_stream(params): + yield x + except ValueError as e: + print("Caught ValueError:", e) + ret = { + "text": server_error_msg, + "error_code": 1, + } + yield json.dumps(ret).encode() + b"\0" + except Exception as e: + print("Caught Unknown Error", e) + ret = { + "text": server_error_msg, + "error_code": 1, + } + yield json.dumps(ret).encode() + b"\0" + + +app = FastAPI() + + +def release_model_semaphore(fn=None): + model_semaphore.release() + if fn is not None: + fn() + + +@app.post("/worker_generate_stream") +async def generate_stream(request: Request): + global model_semaphore, global_counter + global_counter += 1 + params = await request.json() + + if model_semaphore is None: + model_semaphore = asyncio.Semaphore(args.limit_model_concurrency) + await model_semaphore.acquire() + worker.send_heart_beat() + generator = worker.generate_stream_gate(params) + background_tasks = BackgroundTasks() + background_tasks.add_task(partial(release_model_semaphore, fn=worker.send_heart_beat)) + return StreamingResponse(generator, background=background_tasks) + + +@app.post("/worker_get_status") +async def get_status(request: Request): + return worker.get_status() + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + parser.add_argument("--host", type=str, default="localhost") + parser.add_argument("--port", type=int, default=21002) + parser.add_argument("--worker-address", type=str, + default="http://localhost:21002") + parser.add_argument("--controller-address", type=str, + default="http://localhost:21001") + parser.add_argument("--model-name", type=str) + parser.add_argument("--sgl-endpoint", type=str) + parser.add_argument("--limit-model-concurrency", type=int, default=5) + parser.add_argument("--stream-interval", type=int, default=1) + parser.add_argument("--no-register", action="store_true") + args = parser.parse_args() + logger.info(f"args: {args}") + + worker = ModelWorker(args.controller_address, + args.worker_address, + args.sgl_endpoint, + worker_id, + args.no_register, + args.model_name) + uvicorn.run(app, host=args.host, port=args.port, log_level="info") diff --git a/pyproject.toml b/pyproject.toml index 13d4e0144..bc78c8e15 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -14,7 +14,7 @@ classifiers = [ ] dependencies = [ "torch==2.0.1", "torchvision==0.15.2", - "transformers==4.31.0", "tokenizers>=0.12.1,<0.14", "sentencepiece==0.1.99", "shortuuid", + "transformers==4.36.2", "tokenizers==0.15.0", "sentencepiece==0.1.99", "shortuuid", "accelerate==0.21.0", "peft==0.4.0", "bitsandbytes==0.41.0", "pydantic<2,>=1", "markdown2[all]", "numpy", "scikit-learn==1.2.2", "gradio==3.35.2", "gradio_client==0.2.9",