Merge branch 'dev' into feature/lm-studio-prompt-driver

CHANGELOG.md

@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Native function calling support to `OpenAiChatPromptDriver`, `AzureOpenAiChatPromptDriver`, `AnthropicPromptDriver`, `AmazonBedrockPromptDriver`, `GooglePromptDriver`, and `CoherePromptDriver`.
 - `LmStudioPromptDriver` for generating chat completions with LmStudio models.
 - `LmStudioEmbeddingDriver` for generating embeddings with LmStudio models.
+- `OllamaEmbeddingDriver` for generating embeddings with Ollama.
 
 ### Changed
 

docs/griptape-framework/drivers/embedding-drivers.md

@@ -97,8 +97,28 @@ embeddings = driver.embed_string("Hello world!")
 
 # display the first 3 embeddings
 print(embeddings[:3])
+```
+
+### Ollama
+
+!!! info
+    This driver requires the `drivers-embedding-ollama` [extra](../index.md#extras).
+
+The [OllamaEmbeddingDriver](../../reference/griptape/drivers/embedding/ollama_embedding_driver.md) uses the [Ollama Embeddings API](https://ollama.com/blog/embedding-models).
+
+```python title="PYTEST_IGNORE"
+from griptape.drivers import OllamaEmbeddingDriver
+
+driver = OllamaEmbeddingDriver(
+    model="all-minilm",
+)
 
+results = driver.embed_string("Hello world!")
+
+# display the first 3 embeddings
+print(results[:3])
 ```
+
 ### Amazon SageMaker Jumpstart
 
 The [AmazonSageMakerJumpstartEmbeddingDriver](../../reference/griptape/drivers/embedding/amazon_sagemaker_jumpstart_embedding_driver.md) uses the [Amazon SageMaker Endpoints](https://docs.aws.amazon.com/sagemaker/latest/dg/realtime-endpoints.html) to generate embeddings on AWS.

griptape/artifacts/action_artifact.py

@@ -7,23 +7,12 @@
 from griptape.mixins import SerializableMixin
 
 if TYPE_CHECKING:
-    from griptape.common import Action
+    from griptape.common import ToolAction
 
 
 @define()
 class ActionArtifact(BaseArtifact, SerializableMixin):
-    """Represents an instance of an LLM calling a Action.
-
-    Attributes:
-        tag: The tag (unique identifier) of the action.
-        name: The name (Tool name) of the action.
-        path: The path (Tool activity name) of the action.
-        input: The input (Tool params) of the action.
-        tool: The matched Tool of the action.
-        output: The output (Tool result) of the action.
-    """
-
-    value: Action = field(metadata={"serializable": True})
+    value: ToolAction = field(metadata={"serializable": True})
 
     def __add__(self, other: BaseArtifact) -> ActionArtifact:
         raise NotImplementedError

griptape/common/__init__.py

@@ -1,4 +1,5 @@
-from .action import Action
+from .actions.base_action import BaseAction
+from .actions.tool_action import ToolAction
 
 from .prompt_stack.contents.base_message_content import BaseMessageContent
 from .prompt_stack.contents.base_delta_message_content import BaseDeltaMessageContent
@@ -32,5 +33,6 @@
     "ActionResultMessageContent",
     "PromptStack",
     "Reference",
-    "Action",
+    "BaseAction",
+    "ToolAction",
 ]

griptape/common/actions/base_action.py

@@ -0,0 +1,5 @@
+from griptape.mixins import SerializableMixin
+from abc import ABC
+
+
+class BaseAction(SerializableMixin, ABC): ...

griptape/common/action.py → griptape/common/actions/tool_action.py

@@ -5,27 +5,35 @@
 from attrs import define, field
 
 from griptape.artifacts import BaseArtifact
-from griptape.mixins import SerializableMixin
+from griptape.common import BaseAction
 
 if TYPE_CHECKING:
     from griptape.tools import BaseTool
 
 
 @define(kw_only=True)
-class Action(SerializableMixin):
+class ToolAction(BaseAction):
+    """Represents an instance of an LLM using a Tool.
+
+    Attributes:
+        tag: The tag (unique identifier) of the action.
+        name: The name (Tool name) of the action.
+        path: The path (Tool activity name) of the action.
+        input: The input (Tool params) of the action.
+        tool: The matched Tool of the action.
+        output: The output (Tool result) of the action.
+    """
+
     tag: str = field(metadata={"serializable": True})
     name: str = field(metadata={"serializable": True})
     path: Optional[str] = field(default=None, metadata={"serializable": True})
     input: dict = field(factory=dict, metadata={"serializable": True})
     tool: Optional[BaseTool] = field(default=None)
-    output: Optional[BaseArtifact] = field(default=None, metadata={"serializable": True})
+    output: Optional[BaseArtifact] = field(default=None)
 
     def __str__(self) -> str:
         return json.dumps(self.to_dict())
 
-    def to_dict(self) -> dict:
-        return {"tag": self.tag, "name": self.name, "path": self.path, "input": self.input}
-
     def to_native_tool_name(self) -> str:
         parts = [self.name]
 

griptape/common/prompt_stack/contents/action_call_message_content.py

@@ -5,7 +5,7 @@
 
 from attrs import define, field
 
-from griptape.common import Action
+from griptape.common import ToolAction
 from griptape.artifacts import ActionArtifact
 from griptape.common import BaseDeltaMessageContent, BaseMessageContent, ActionCallDeltaMessageContent
 
@@ -37,10 +37,10 @@ def from_deltas(cls, deltas: Sequence[BaseDeltaMessageContent]) -> ActionCallMes
             try:
                 parsed_input = json.loads(input)
             except json.JSONDecodeError as exc:
-                raise ValueError("Invalid JSON input for Action") from exc
-            action = Action(tag=tag, name=name, path=path, input=parsed_input)
+                raise ValueError("Invalid JSON input for ToolAction") from exc
+            action = ToolAction(tag=tag, name=name, path=path, input=parsed_input)
         else:
-            raise ValueError("Missing required fields for Action")
+            raise ValueError("Missing required fields for ToolAction")
 
         artifact = ActionArtifact(value=action)
 

griptape/common/prompt_stack/contents/action_result_message_content.py

@@ -5,13 +5,13 @@
 from attrs import define, field
 
 from griptape.artifacts import BaseArtifact
-from griptape.common import BaseDeltaMessageContent, BaseMessageContent, Action
+from griptape.common import BaseDeltaMessageContent, BaseMessageContent, ToolAction
 
 
 @define
 class ActionResultMessageContent(BaseMessageContent):
     artifact: BaseArtifact = field(metadata={"serializable": True})
-    action: Action = field(metadata={"serializable": True})
+    action: ToolAction = field(metadata={"serializable": True})
 
     @classmethod
     def from_deltas(cls, deltas: Sequence[BaseDeltaMessageContent]) -> ActionResultMessageContent:

griptape/drivers/__init__.py

@@ -29,6 +29,7 @@
 from .embedding.dummy_embedding_driver import DummyEmbeddingDriver
 from .embedding.cohere_embedding_driver import CohereEmbeddingDriver
 from .embedding.lm_studio_embedding_driver import LmStudioEmbeddingDriver
+from .embedding.ollama_embedding_driver import OllamaEmbeddingDriver
 
 from .vector.base_vector_store_driver import BaseVectorStoreDriver
 from .vector.local_vector_store_driver import LocalVectorStoreDriver
@@ -139,6 +140,7 @@
     "DummyEmbeddingDriver",
     "CohereEmbeddingDriver",
     "LmStudioEmbeddingDriver",
+    "OllamaEmbeddingDriver",
     "BaseVectorStoreDriver",
     "LocalVectorStoreDriver",
     "PineconeVectorStoreDriver",

griptape/drivers/embedding/ollama_embedding_driver.py

@@ -0,0 +1,28 @@
+from __future__ import annotations
+from typing import Optional, TYPE_CHECKING
+from attrs import define, field, Factory
+from griptape.utils import import_optional_dependency
+from griptape.drivers import BaseEmbeddingDriver
+
+if TYPE_CHECKING:
+    from ollama import Client
+
+
+@define
+class OllamaEmbeddingDriver(BaseEmbeddingDriver):
+    """
+    Attributes:
+        model: Ollama embedding model name.
+        host: Optional Ollama host.
+        client: Ollama `Client`.
+    """
+
+    model: str = field(kw_only=True, metadata={"serializable": True})
+    host: Optional[str] = field(default=None, kw_only=True, metadata={"serializable": True})
+    client: Client = field(
+        default=Factory(lambda self: import_optional_dependency("ollama").Client(host=self.host), takes_self=True),
+        kw_only=True,
+    )
+
+    def try_embed_chunk(self, chunk: str) -> list[float]:
+        return list(self.client.embeddings(model=self.model, prompt=chunk)["embedding"])

griptape/drivers/prompt/amazon_bedrock_prompt_driver.py

@@ -26,7 +26,7 @@
     Message,
     TextDeltaMessageContent,
     TextMessageContent,
-    Action,
+    ToolAction,
 )
 from griptape.drivers import BasePromptDriver
 from griptape.tokenizers import AmazonBedrockTokenizer, BaseTokenizer
@@ -177,10 +177,10 @@ def __to_prompt_stack_message_content(self, content: dict) -> BaseMessageContent
         if "text" in content:
             return TextMessageContent(TextArtifact(content["text"]))
         elif "toolUse" in content:
-            name, path = Action.from_native_tool_name(content["toolUse"]["name"])
+            name, path = ToolAction.from_native_tool_name(content["toolUse"]["name"])
             return ActionCallMessageContent(
                 artifact=ActionArtifact(
-                    value=Action(
+                    value=ToolAction(
                         tag=content["toolUse"]["toolUseId"], name=name, path=path, input=content["toolUse"]["input"]
                     )
                 )
@@ -193,7 +193,7 @@ def __to_prompt_stack_delta_message_content(self, event: dict) -> BaseDeltaMessa
             content_block = event["contentBlockStart"]["start"]
 
             if "toolUse" in content_block:
-                name, path = Action.from_native_tool_name(content_block["toolUse"]["name"])
+                name, path = ToolAction.from_native_tool_name(content_block["toolUse"]["name"])
 
                 return ActionCallDeltaMessageContent(
                     index=event["contentBlockStart"]["contentBlockIndex"],

griptape/drivers/prompt/anthropic_prompt_driver.py

@@ -26,7 +26,7 @@
     ImageMessageContent,
     PromptStack,
     Message,
-    Action,
+    ToolAction,
     TextMessageContent,
 )
 from griptape.drivers import BasePromptDriver
@@ -183,11 +183,11 @@ def __to_prompt_stack_message_content(self, content: ContentBlock) -> BaseMessag
         if content.type == "text":
             return TextMessageContent(TextArtifact(content.text))
         elif content.type == "tool_use":
-            name, path = Action.from_native_tool_name(content.name)
+            name, path = ToolAction.from_native_tool_name(content.name)
 
             return ActionCallMessageContent(
                 artifact=ActionArtifact(
-                    value=Action(tag=content.id, name=name, path=path, input=content.input)  # pyright: ignore[reportArgumentType]
+                    value=ToolAction(tag=content.id, name=name, path=path, input=content.input)  # pyright: ignore[reportArgumentType]
                 )
             )
         else:
@@ -200,7 +200,7 @@ def __to_prompt_stack_delta_message_content(
             content_block = event.content_block
 
             if content_block.type == "tool_use":
-                name, path = Action.from_native_tool_name(content_block.name)
+                name, path = ToolAction.from_native_tool_name(content_block.name)
 
                 return ActionCallDeltaMessageContent(index=event.index, tag=content_block.id, name=name, path=path)
             elif content_block.type == "text":

griptape/drivers/prompt/cohere_prompt_driver.py

@@ -18,7 +18,7 @@
     Message,
     TextMessageContent,
     ActionResultMessageContent,
-    Action,
+    ToolAction,
 )
 from griptape.utils import import_optional_dependency
 from griptape.tokenizers import BaseTokenizer
@@ -208,10 +208,10 @@ def __to_prompt_stack_message_content(self, response: NonStreamedChatResponse) -
                 [
                     ActionCallMessageContent(
                         ActionArtifact(
-                            Action(
+                            ToolAction(
                                 tag=tool_call.name,
-                                name=Action.from_native_tool_name(tool_call.name)[0],
-                                path=Action.from_native_tool_name(tool_call.name)[1],
+                                name=ToolAction.from_native_tool_name(tool_call.name)[0],
+                                path=ToolAction.from_native_tool_name(tool_call.name)[1],
                                 input=tool_call.parameters,
                             )
                         )
@@ -229,7 +229,7 @@ def __to_prompt_stack_delta_message_content(self, event: Any) -> BaseDeltaMessag
             if event.tool_call_delta is not None:
                 tool_call_delta = event.tool_call_delta
                 if tool_call_delta.name is not None:
-                    name, path = Action.from_native_tool_name(tool_call_delta.name)
+                    name, path = ToolAction.from_native_tool_name(tool_call_delta.name)
 
                     return ActionCallDeltaMessageContent(tag=tool_call_delta.name, name=name, path=path)
                 else:

griptape/drivers/prompt/google_prompt_driver.py

@@ -20,7 +20,7 @@
     ActionResultMessageContent,
     ActionCallDeltaMessageContent,
     BaseDeltaMessageContent,
-    Action,
+    ToolAction,
 )
 from griptape.artifacts import TextArtifact, ActionArtifact
 from griptape.drivers import BasePromptDriver
@@ -219,11 +219,11 @@ def __to_prompt_stack_message_content(self, content: Part) -> BaseMessageContent
         elif content.function_call:
             function_call = content.function_call
 
-            name, path = Action.from_native_tool_name(function_call.name)
+            name, path = ToolAction.from_native_tool_name(function_call.name)
 
             args = {k: v for k, v in function_call.args.items()}
             return ActionCallMessageContent(
-                artifact=ActionArtifact(value=Action(tag=function_call.name, name=name, path=path, input=args))
+                artifact=ActionArtifact(value=ToolAction(tag=function_call.name, name=name, path=path, input=args))
             )
         else:
             raise ValueError(f"Unsupported message content type {content}")
@@ -234,7 +234,7 @@ def __to_prompt_stack_delta_message_content(self, content: Part) -> BaseDeltaMes
         elif content.function_call:
             function_call = content.function_call
 
-            name, path = Action.from_native_tool_name(function_call.name)
+            name, path = ToolAction.from_native_tool_name(function_call.name)
 
             args = {k: v for k, v in function_call.args.items()}
             return ActionCallDeltaMessageContent(

griptape/drivers/prompt/openai_chat_prompt_driver.py

@@ -21,7 +21,7 @@
     PromptStack,
     Message,
     TextMessageContent,
-    Action,
+    ToolAction,
 )
 from griptape.drivers import BasePromptDriver
 from griptape.tokenizers import BaseTokenizer, OpenAiTokenizer
@@ -153,7 +153,7 @@ def __to_openai_messages(self, messages: list[Message]) -> list[dict]:
                 if message.to_text():
                     openai_messages.append({"role": message.role, "content": message.to_text()})
             elif message.has_any_content_type(ActionResultMessageContent):
-                # Action results need to be expanded into separate messages.
+                # ToolAction results need to be expanded into separate messages.
                 openai_messages.extend(
                     [
                         {
@@ -165,7 +165,7 @@ def __to_openai_messages(self, messages: list[Message]) -> list[dict]:
                     ]
                 )
             else:
-                # Action calls are attached to the assistant message that originally generated them.
+                # ToolAction calls are attached to the assistant message that originally generated them.
                 action_call_content = []
                 non_action_call_content = []
                 for content in message.content:
@@ -179,7 +179,7 @@ def __to_openai_messages(self, messages: list[Message]) -> list[dict]:
                         "role": self.__to_openai_role(message),
                         "content": [
                             self.__to_openai_message_content(content)
-                            for content in non_action_call_content  # Action calls do not belong in the content
+                            for content in non_action_call_content  # ToolAction calls do not belong in the content
                         ],
                         **(
                             {
@@ -251,10 +251,10 @@ def __to_prompt_stack_message_content(self, response: ChatCompletionMessage) ->
                 [
                     ActionCallMessageContent(
                         ActionArtifact(
-                            Action(
+                            ToolAction(
                                 tag=tool_call.id,
-                                name=Action.from_native_tool_name(tool_call.function.name)[0],
-                                path=Action.from_native_tool_name(tool_call.function.name)[1],
+                                name=ToolAction.from_native_tool_name(tool_call.function.name)[0],
+                                path=ToolAction.from_native_tool_name(tool_call.function.name)[1],
                                 input=json.loads(tool_call.function.arguments),
                             )
                         )
@@ -280,8 +280,8 @@ def __to_prompt_stack_delta_message_content(self, content_delta: ChoiceDelta) ->
                     return ActionCallDeltaMessageContent(
                         index=index,
                         tag=tool_call.id,
-                        name=Action.from_native_tool_name(tool_call.function.name)[0],
-                        path=Action.from_native_tool_name(tool_call.function.name)[1],
+                        name=ToolAction.from_native_tool_name(tool_call.function.name)[0],
+                        path=ToolAction.from_native_tool_name(tool_call.function.name)[1],
                     )
                 else:
                     return ActionCallDeltaMessageContent(index=index, partial_input=tool_call.function.arguments)