Merge branch 'main' into feature/datastore

app/llm/__init__.py

@@ -1,3 +1,3 @@
-from llm.request_handler_interface import RequestHandlerInterface
-from llm.generation_arguments import *
-from llm.basic_request_handler import BasicRequestHandler, BasicRequestHandlerModel
+from llm.request_handler_interface import RequestHandler
+from llm.completion_arguments import *
+from llm.basic_request_handler import BasicRequestHandler, DefaultModelId

app/llm/basic_request_handler.py

@@ -1,48 +1,26 @@
 from domain import IrisMessage
-from llm import RequestHandlerInterface, CompletionArguments
+from llm import RequestHandler, CompletionArguments
 from llm.llm_manager import LlmManager
-from llm.wrapper.abstract_llm_wrapper import (
-    AbstractLlmCompletionWrapper,
-    AbstractLlmChatCompletionWrapper,
-    AbstractLlmEmbeddingWrapper,
-)
 
-type BasicRequestHandlerModel = str
 
-
-class BasicRequestHandler(RequestHandlerInterface):
-    model: BasicRequestHandlerModel
+class BasicRequestHandler(RequestHandler):
+    model_id: str
     llm_manager: LlmManager
 
-    def __init__(self, model: BasicRequestHandlerModel):
-        self.model = model
+    def __init__(self, model_id: str):
+        self.model_id = model_id
         self.llm_manager = LlmManager()
 
-    def completion(self, prompt: str, arguments: CompletionArguments) -> str:
-        llm = self.llm_manager.get_llm_by_id(self.model).llm
-        if isinstance(llm, AbstractLlmCompletionWrapper):
-            return llm.completion(prompt, arguments)
-        else:
-            raise NotImplementedError(
-                f"The LLM {llm.__str__()} does not support completion"
-            )
+    def complete(self, prompt: str, arguments: CompletionArguments) -> str:
+        llm = self.llm_manager.get_by_id(self.model_id)
+        return llm.complete(prompt, arguments)
 
-    def chat_completion(
+    def chat(
         self, messages: list[IrisMessage], arguments: CompletionArguments
     ) -> IrisMessage:
-        llm = self.llm_manager.get_llm_by_id(self.model).llm
-        if isinstance(llm, AbstractLlmChatCompletionWrapper):
-            return llm.chat_completion(messages, arguments)
-        else:
-            raise NotImplementedError(
-                f"The LLM {llm.__str__()} does not support chat completion"
-            )
+        llm = self.llm_manager.get_by_id(self.model_id)
+        return llm.chat(messages, arguments)
 
-    def create_embedding(self, text: str) -> list[float]:
-        llm = self.llm_manager.get_llm_by_id(self.model).llm
-        if isinstance(llm, AbstractLlmEmbeddingWrapper):
-            return llm.create_embedding(text)
-        else:
-            raise NotImplementedError(
-                f"The LLM {llm.__str__()} does not support embedding"
-            )
+    def embed(self, text: str) -> list[float]:
+        llm = self.llm_manager.get_by_id(self.model_id)
+        return llm.embed(text)

app/llm/external/__init__.py

@@ -0,0 +1,21 @@
+from llm.external.model import LanguageModel
+from llm.external.openai_completion import (
+    DirectOpenAICompletionModel,
+    AzureOpenAICompletionModel,
+)
+from llm.external.openai_chat import DirectOpenAIChatModel, AzureOpenAIChatModel
+from llm.external.openai_embeddings import (
+    DirectOpenAIEmbeddingModel,
+    AzureOpenAIEmbeddingModel,
+)
+from llm.external.ollama import OllamaModel
+
+type AnyLLM = (
+    DirectOpenAICompletionModel
+    | AzureOpenAICompletionModel
+    | DirectOpenAIChatModel
+    | AzureOpenAIChatModel
+    | DirectOpenAIEmbeddingModel
+    | AzureOpenAIEmbeddingModel
+    | OllamaModel
+)

app/llm/external/model.py

@@ -0,0 +1,60 @@
+from abc import ABCMeta, abstractmethod
+from pydantic import BaseModel
+
+from domain import IrisMessage
+from llm import CompletionArguments
+
+
+class LanguageModel(BaseModel, metaclass=ABCMeta):
+    """Abstract class for the llm wrappers"""
+
+    id: str
+    name: str
+    description: str
+
+
+class CompletionModel(LanguageModel, metaclass=ABCMeta):
+    """Abstract class for the llm completion wrappers"""
+
+    @classmethod
+    def __subclasshook__(cls, subclass) -> bool:
+        return hasattr(subclass, "complete") and callable(subclass.complete)
+
+    @abstractmethod
+    def complete(self, prompt: str, arguments: CompletionArguments) -> str:
+        """Create a completion from the prompt"""
+        raise NotImplementedError(
+            f"The LLM {self.__str__()} does not support completion"
+        )
+
+
+class ChatModel(LanguageModel, metaclass=ABCMeta):
+    """Abstract class for the llm chat completion wrappers"""
+
+    @classmethod
+    def __subclasshook__(cls, subclass) -> bool:
+        return hasattr(subclass, "chat") and callable(subclass.chat)
+
+    @abstractmethod
+    def chat(
+        self, messages: list[IrisMessage], arguments: CompletionArguments
+    ) -> IrisMessage:
+        """Create a completion from the chat messages"""
+        raise NotImplementedError(
+            f"The LLM {self.__str__()} does not support chat completion"
+        )
+
+
+class EmbeddingModel(LanguageModel, metaclass=ABCMeta):
+    """Abstract class for the llm embedding wrappers"""
+
+    @classmethod
+    def __subclasshook__(cls, subclass) -> bool:
+        return hasattr(subclass, "embed") and callable(subclass.embed)
+
+    @abstractmethod
+    def embed(self, text: str) -> list[float]:
+        """Create an embedding from the text"""
+        raise NotImplementedError(
+            f"The LLM {self.__str__()} does not support embeddings"
+        )

app/llm/wrapper/ollama_wrapper.py → app/llm/external/ollama.py

@@ -4,11 +4,7 @@
 
 from domain import IrisMessage, IrisMessageRole
 from llm import CompletionArguments
-from llm.wrapper.abstract_llm_wrapper import (
-    AbstractLlmChatCompletionWrapper,
-    AbstractLlmCompletionWrapper,
-    AbstractLlmEmbeddingWrapper,
-)
+from llm.external.model import ChatModel, CompletionModel, EmbeddingModel
 
 
 def convert_to_ollama_messages(messages: list[IrisMessage]) -> list[Message]:
@@ -21,10 +17,10 @@ def convert_to_iris_message(message: Message) -> IrisMessage:
     return IrisMessage(role=IrisMessageRole(message["role"]), text=message["content"])
 
 
-class OllamaWrapper(
-    AbstractLlmCompletionWrapper,
-    AbstractLlmChatCompletionWrapper,
-    AbstractLlmEmbeddingWrapper,
+class OllamaModel(
+    CompletionModel,
+    ChatModel,
+    EmbeddingModel,
 ):
     type: Literal["ollama"]
     model: str
@@ -34,19 +30,19 @@ class OllamaWrapper(
     def model_post_init(self, __context: Any) -> None:
         self._client = Client(host=self.host)  # TODO: Add authentication (httpx auth?)
 
-    def completion(self, prompt: str, arguments: CompletionArguments) -> str:
+    def complete(self, prompt: str, arguments: CompletionArguments) -> str:
         response = self._client.generate(model=self.model, prompt=prompt)
         return response["response"]
 
-    def chat_completion(
-        self, messages: list[any], arguments: CompletionArguments
-    ) -> any:
+    def chat(
+        self, messages: list[IrisMessage], arguments: CompletionArguments
+    ) -> IrisMessage:
         response = self._client.chat(
             model=self.model, messages=convert_to_ollama_messages(messages)
         )
         return convert_to_iris_message(response["message"])
 
-    def create_embedding(self, text: str) -> list[float]:
+    def embed(self, text: str) -> list[float]:
         response = self._client.embeddings(model=self.model, prompt=text)
         return list(response)
 

app/llm/wrapper/open_ai_chat_wrapper.py → app/llm/external/openai_chat.py

@@ -1,11 +1,12 @@
 from typing import Literal, Any
+
 from openai import OpenAI
 from openai.lib.azure import AzureOpenAI
 from openai.types.chat import ChatCompletionMessageParam, ChatCompletionMessage
 
 from domain import IrisMessage, IrisMessageRole
 from llm import CompletionArguments
-from llm.wrapper.abstract_llm_wrapper import AbstractLlmChatCompletionWrapper
+from llm.external.model import ChatModel
 
 
 def convert_to_open_ai_messages(
@@ -22,13 +23,13 @@ def convert_to_iris_message(message: ChatCompletionMessage) -> IrisMessage:
     return IrisMessage(role=message_role, text=message.content)
 
 
-class BaseOpenAIChatCompletionWrapper(AbstractLlmChatCompletionWrapper):
+class OpenAIChatModel(ChatModel):
     model: str
     api_key: str
     _client: OpenAI
 
-    def chat_completion(
-        self, messages: list[any], arguments: CompletionArguments
+    def chat(
+        self, messages: list[IrisMessage], arguments: CompletionArguments
     ) -> IrisMessage:
         response = self._client.chat.completions.create(
             model=self.model,
@@ -40,7 +41,7 @@ def chat_completion(
         return convert_to_iris_message(response.choices[0].message)
 
 
-class OpenAIChatCompletionWrapper(BaseOpenAIChatCompletionWrapper):
+class DirectOpenAIChatModel(OpenAIChatModel):
     type: Literal["openai_chat"]
 
     def model_post_init(self, __context: Any) -> None:
@@ -50,7 +51,7 @@ def __str__(self):
         return f"OpenAIChat('{self.model}')"
 
 
-class AzureChatCompletionWrapper(BaseOpenAIChatCompletionWrapper):
+class AzureOpenAIChatModel(OpenAIChatModel):
     type: Literal["azure_chat"]
     endpoint: str
     azure_deployment: str

app/llm/wrapper/open_ai_completion_wrapper.py → app/llm/external/openai_completion.py

@@ -3,15 +3,15 @@
 from openai.lib.azure import AzureOpenAI
 
 from llm import CompletionArguments
-from llm.wrapper.abstract_llm_wrapper import AbstractLlmCompletionWrapper
+from llm.external.model import CompletionModel
 
 
-class BaseOpenAICompletionWrapper(AbstractLlmCompletionWrapper):
+class OpenAICompletionModel(CompletionModel):
     model: str
     api_key: str
     _client: OpenAI
 
-    def completion(self, prompt: str, arguments: CompletionArguments) -> any:
+    def complete(self, prompt: str, arguments: CompletionArguments) -> any:
         response = self._client.completions.create(
             model=self.model,
             prompt=prompt,
@@ -22,7 +22,7 @@ def completion(self, prompt: str, arguments: CompletionArguments) -> any:
         return response
 
 
-class OpenAICompletionWrapper(BaseOpenAICompletionWrapper):
+class DirectOpenAICompletionModel(OpenAICompletionModel):
     type: Literal["openai_completion"]
 
     def model_post_init(self, __context: Any) -> None:
@@ -32,7 +32,7 @@ def __str__(self):
         return f"OpenAICompletion('{self.model}')"
 
 
-class AzureCompletionWrapper(BaseOpenAICompletionWrapper):
+class AzureOpenAICompletionModel(OpenAICompletionModel):
     type: Literal["azure_completion"]
     endpoint: str
     azure_deployment: str

app/llm/wrapper/open_ai_embedding_wrapper.py → app/llm/external/openai_embeddings.py

@@ -2,15 +2,15 @@
 from openai import OpenAI
 from openai.lib.azure import AzureOpenAI
 
-from llm.wrapper.abstract_llm_wrapper import AbstractLlmEmbeddingWrapper
+from llm.external.model import EmbeddingModel
 
 
-class BaseOpenAIEmbeddingWrapper(AbstractLlmEmbeddingWrapper):
+class OpenAIEmbeddingModel(EmbeddingModel):
     model: str
     api_key: str
     _client: OpenAI
 
-    def create_embedding(self, text: str) -> list[float]:
+    def embed(self, text: str) -> list[float]:
         response = self._client.embeddings.create(
             model=self.model,
             input=text,
@@ -19,7 +19,7 @@ def create_embedding(self, text: str) -> list[float]:
         return response.data[0].embedding
 
 
-class OpenAIEmbeddingWrapper(BaseOpenAIEmbeddingWrapper):
+class DirectOpenAIEmbeddingModel(OpenAIEmbeddingModel):
     type: Literal["openai_embedding"]
 
     def model_post_init(self, __context: Any) -> None:
@@ -29,7 +29,7 @@ def __str__(self):
         return f"OpenAIEmbedding('{self.model}')"
 
 
-class AzureEmbeddingWrapper(BaseOpenAIEmbeddingWrapper):
+class AzureOpenAIEmbeddingModel(OpenAIEmbeddingModel):
     type: Literal["azure_embedding"]
     endpoint: str
     azure_deployment: str

app/llm/langchain/__init__.py

@@ -0,0 +1,3 @@
+from llm.langchain.iris_langchain_completion_model import IrisLangchainCompletionModel
+from llm.langchain.iris_langchain_chat_model import IrisLangchainChatModel
+from llm.langchain.iris_langchain_embedding_model import IrisLangchainEmbeddingModel

app/llm/langchain/iris_langchain_chat_model.py

@@ -0,0 +1,60 @@
+from typing import List, Optional, Any
+
+from langchain_core.callbacks import CallbackManagerForLLMRun
+from langchain_core.language_models.chat_models import (
+    BaseChatModel,
+)
+from langchain_core.messages import BaseMessage
+from langchain_core.outputs import ChatResult
+from langchain_core.outputs.chat_generation import ChatGeneration
+
+from domain import IrisMessage, IrisMessageRole
+from llm import RequestHandler, CompletionArguments
+
+
+def convert_iris_message_to_base_message(iris_message: IrisMessage) -> BaseMessage:
+    role_map = {
+        IrisMessageRole.USER: "human",
+        IrisMessageRole.ASSISTANT: "ai",
+        IrisMessageRole.SYSTEM: "system",
+    }
+    return BaseMessage(content=iris_message.text, type=role_map[iris_message.role])
+
+
+def convert_base_message_to_iris_message(base_message: BaseMessage) -> IrisMessage:
+    role_map = {
+        "human": IrisMessageRole.USER,
+        "ai": IrisMessageRole.ASSISTANT,
+        "system": IrisMessageRole.SYSTEM,
+    }
+    return IrisMessage(
+        text=base_message.content, role=IrisMessageRole(role_map[base_message.type])
+    )
+
+
+class IrisLangchainChatModel(BaseChatModel):
+    """Custom langchain chat model for our own request handler"""
+
+    request_handler: RequestHandler
+
+    def __init__(self, request_handler: RequestHandler, **kwargs: Any) -> None:
+        super().__init__(request_handler=request_handler, **kwargs)
+
+    def _generate(
+        self,
+        messages: List[BaseMessage],
+        stop: Optional[List[str]] = None,
+        run_manager: Optional[CallbackManagerForLLMRun] = None,
+        **kwargs: Any
+    ) -> ChatResult:
+        iris_messages = [convert_base_message_to_iris_message(m) for m in messages]
+        iris_message = self.request_handler.chat(
+            iris_messages, CompletionArguments(stop=stop)
+        )
+        base_message = convert_iris_message_to_base_message(iris_message)
+        chat_generation = ChatGeneration(message=base_message)
+        return ChatResult(generations=[chat_generation])
+
+    @property
+    def _llm_type(self) -> str:
+        return "Iris"