Pull request #1: Feature/init

Merge in SE471652/llm_factory from feature/init to develop

* commit '1dfedfb03bbe9cbf59aba493f2b899b00af6ec99':
  basic initialization test added
  base_query_engine added
  errors added
  llms added
  query engine added
  requirements file created
  readme file created
  toml file created

pyproject.toml

@@ -0,0 +1,75 @@
+[project]
+name = "llm-factory"
+version = "0.0.1"
+authors = [
+  { name="Mithat Sinan Ergen", email="mithat.ergen@turkcell.com.tr" },
+]
+description = "LLM Factory package"
+readme = "README.md"
+requires-python = ">=3.9"
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "Operating System :: OS Independent",
+]
+
+[build-system]
+requires = ["setuptools>=61.0",
+            "build==0.9.0",
+            "pep517==0.13.0",
+            "pyparsing==3.0.9",
+            "tomli==2.0.1",
+            "colorama==0.4.6",
+            "greenlet==1.1.3.post0",
+             "aiohappyeyeballs==2.4.0",
+            "aiohttp==3.10.5",
+            "aiosignal==1.3.1",
+            "annotated-types==0.7.0",
+            "anyio==4.4.0",
+            "async-timeout==4.0.3",
+            "attrs==24.2.0",
+            "certifi==2024.8.30",
+            "charset-normalizer==3.3.2",
+            "dataclasses-json==0.6.7",
+            "distro==1.9.0",
+            "exceptiongroup==1.2.2",
+            "frozenlist==1.4.1",
+            "h11==0.14.0",
+            "httpcore==1.0.5",
+            "httpx==0.27.2",
+            "idna==3.8",
+            "jiter==0.5.0",
+            "jsonpatch==1.33",
+            "jsonpointer==3.0.0",
+            "langchain==0.2.6",
+            "langchain-community==0.2.6",
+            "langchain-core==0.2.11",
+            "langchain-openai==0.1.14",
+            "langchain-text-splitters==0.2.2",
+            "langsmith==0.1.116",
+            "loguru==0.7.2",
+            "marshmallow==3.22.0",
+            "multidict==6.0.5",
+            "mypy-extensions==1.0.0",
+            "numpy==1.26.4",
+            "openai==1.44.0",
+            "orjson==3.10.7",
+            "packaging==24.1",
+            "pydantic==2.9.0",
+            "pydantic_core==2.23.2",
+            "PyYAML==6.0.2",
+            "regex==2024.7.24",
+            "requests==2.32.3",
+            "sniffio==1.3.1",
+            "SQLAlchemy==2.0.34",
+            "tenacity==8.5.0",
+            "tiktoken==0.7.0",
+            "tqdm==4.66.5",
+            "typing-inspect==0.9.0",
+            "typing_extensions==4.12.2",
+            "tzdata==2024.1",
+            "urllib3==2.2.2",
+            "yarl==1.11.0",
+            "win32-setctime==1.1.0"]
+build-backend = "setuptools.build_meta"
+
+

requirements.txt

@@ -0,0 +1,49 @@
+aiohappyeyeballs==2.4.0
+aiohttp==3.10.5
+aiosignal==1.3.1
+annotated-types==0.7.0
+anyio==4.4.0
+async-timeout==4.0.3
+attrs==24.2.0
+certifi==2024.8.30
+charset-normalizer==3.3.2
+dataclasses-json==0.6.7
+distro==1.9.0
+exceptiongroup==1.2.2
+frozenlist==1.4.1
+h11==0.14.0
+httpcore==1.0.5
+httpx==0.27.2
+idna==3.8
+jiter==0.5.0
+jsonpatch==1.33
+jsonpointer==3.0.0
+langchain==0.2.6
+langchain-community==0.2.6
+langchain-core==0.2.11
+langchain-openai==0.1.14
+langchain-text-splitters==0.2.2
+langsmith==0.1.116
+loguru==0.7.2
+marshmallow==3.22.0
+multidict==6.0.5
+mypy-extensions==1.0.0
+numpy==1.26.4
+openai==1.44.0
+orjson==3.10.7
+packaging==24.1
+pydantic==2.9.0
+pydantic_core==2.23.2
+PyYAML==6.0.2
+regex==2024.7.24
+requests==2.32.3
+sniffio==1.3.1
+SQLAlchemy==2.0.34
+tenacity==8.5.0
+tiktoken==0.7.0
+tqdm==4.66.5
+typing-inspect==0.9.0
+typing_extensions==4.12.2
+tzdata==2024.1
+urllib3==2.2.2
+yarl==1.11.0

src/base_query_engine.py

@@ -0,0 +1,45 @@
+from abc import ABC, abstractmethod
+from langchain_community.callbacks import get_openai_callback
+from loguru import logger
+
+
+class BaseQueryEngine(ABC):
+
+    def __init__(self, model, provider):
+        self.model = model
+        self.provider = provider
+
+    @abstractmethod
+    def get_sample_message_content(self):
+        pass
+
+    @abstractmethod
+    def construct_message_content(self, inputs):
+        pass
+
+    @abstractmethod
+    def calculate_consumption(self, cb):
+        pass
+
+    def get_provider_name(self):
+        return self.provider
+
+    def ask(self, inputs):
+        with get_openai_callback() as cb:
+            message_content = self.construct_message_content(inputs)
+            response = self.model.invoke(message_content)
+
+            logger.debug(f"Total Tokens: {cb.total_tokens}")
+            logger.debug(f"Prompt Tokens: {cb.prompt_tokens}")
+            logger.debug(f"Completion Tokens: {cb.completion_tokens}")
+            logger.debug(f"Total Cost (USD): ${round(cb.total_cost, 4)}")
+
+            consumption = self.calculate_consumption(cb)
+
+        answer = self.parse_response(response, inputs)
+        return answer, consumption
+
+    def parse_response(self, response, inputs):
+        if "parser" in inputs:
+            return inputs["parser"].invoke(input=response.content)
+        return response.content

src/errors.py

@@ -0,0 +1,25 @@
+
+
+class InsertionError(Exception):
+    def __init__(self, message):
+        super().__init__(message)
+
+
+class DeletionError(Exception):
+    def __init__(self, message):
+        super().__init__(message)
+
+
+class LanguageError(Exception):
+    def __init__(self, message):
+        super().__init__(message)
+
+
+class LLMProviderError(ConnectionError):
+    def __init__(self, message):
+        super().__init__(message)
+
+
+class LLMFactoryError(ConnectionError):
+    def __init__(self, message):
+        super().__init__(message)

src/llms.py

@@ -0,0 +1,84 @@
+import os
+import query_engine as qe
+
+from typing import List
+
+from errors import LLMFactoryError
+from langchain_openai import ChatOpenAI, AzureChatOpenAI
+from langchain_community.chat_models import ChatOllama
+
+
+class LLMFactory:
+
+    def __init__(self):
+        self.model = None
+
+    def create_model(self, provider: str, conversational: bool, model_name: str, multimodal: List[str] = None):
+        if provider == "azure" and conversational:
+            self.model = AzureChatOpenAI(
+                openai_api_key=os.environ["AZURE_OPENAI_API_KEY"],
+                openai_api_version=os.environ["AZURE_API_VERSION"],
+                azure_endpoint=os.environ["AZURE_OPENAI_API_ENDPOINT"],
+                azure_deployment=model_name,
+            )
+            return qe.QueryEngineOpenAIConversationalText(self.model)
+        elif provider == "openai":
+            if conversational:
+                if multimodal:
+                    if multimodal == ["text", "image"]:
+                        self.model = ChatOpenAI(
+                            temperature=os.environ["OPENAI_TEMPERATURE"],
+                            model=model_name,
+                            timeout=10,
+                            max_retries=3
+                        )
+                        return qe.QueryEngineOpenAIConversationalMultiModal(self.model)
+                    else:
+                        raise LLMFactoryError(f"given multimodality is not implemented: {multimodal}")
+                else:
+                    raise LLMFactoryError(f"non-multimodal is not implemented")
+            else:
+                raise LLMFactoryError(f"non-conversation is not implemented")
+        elif provider == "ollama":
+            if conversational:
+                self.model = ChatOllama(
+                    base_url=os.environ["OLLAMA_BASE_URL"],
+                    temperature=0,
+                    model=model_name,
+                    keep_alive=int(os.environ["OLLAMA_KEEP_ALIVE"]),
+                    num_ctx=int(os.environ["OLLAMA_NUM_CTX"]),
+                    timeout=300,
+                    max_retries=3
+                )
+                if multimodal:
+                    if multimodal == ["text", "image"]:
+                        return qe.QueryEngineOllamaConversationalMultiModal(self.model, provider)
+                    else:
+                        raise LLMFactoryError(f"given multimodality is not implemented: {multimodal}")
+                else:
+                    return qe.QueryEngineOllamaConversationalText(self.model, provider)
+            else:
+                raise LLMFactoryError(f"non-conversation is not implemented")
+        elif provider == "llamacpp":
+            if conversational:
+                self.model = ChatOpenAI(
+                    model=model_name, #"models/Meta-Llama-3-8B-Instruct.Q8_0.gguf",
+                    temperature=0,
+                    max_tokens=2048,
+                    stop=["<|eot_id|>"],
+                    timeout=None,
+                    max_retries=2,
+                    api_key="sk-xxx",
+                    base_url=os.environ["LLAMACPP_BASE_URL"],
+                )
+                if multimodal:
+                    if multimodal == ["text", "image"]:
+                        raise LLMFactoryError(f"given {provider}, multimodality is not implemented: {multimodal}")
+                    else:
+                        raise LLMFactoryError(f"given {provider}, multimodality is not implemented: {multimodal}")
+                else:
+                    return qe.QueryEngineLlamacppConversationalText(self.model)
+            else:
+                raise LLMFactoryError(f"non-conversation is not implemented")
+
+

src/query_engine.py

@@ -0,0 +1,92 @@
+from langchain_community.callbacks import get_openai_callback
+from loguru import logger
+
+from base_query_engine import BaseQueryEngine
+from langchain_core.messages import HumanMessage
+
+
+class QueryEngineOpenAIConversationalMultiModal(BaseQueryEngine):
+
+    def construct_message_content(self, inputs):
+        return [
+            HumanMessage(content=[
+                {"type": "text", "text": inputs["prompt"]},
+                {"type": "text", "text": inputs["parser"].get_format_instructions()},
+                {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{inputs['image']}"}}
+            ])
+        ]
+
+    def calculate_consumption(self, cb):
+        return {"total_cost": round(cb.total_cost, 4), "total_tokens": cb.total_tokens}
+
+
+class QueryEngineOpenAIConversationalText(BaseQueryEngine):
+
+    def construct_message_content(self, inputs):
+        return [
+            HumanMessage(content=[
+                {"type": "text", "text": inputs["prompt"]}
+            ])
+        ]
+
+    def calculate_consumption(self, cb):
+        return {"total_cost": round(cb.total_cost, 4), "total_tokens": cb.total_tokens}
+
+
+class QueryEngineOllamaConversationalMultiModal(BaseQueryEngine):
+
+    def construct_message_content(self, inputs):
+        return [
+            HumanMessage(content=[
+                {"type": "text", "text": inputs["prompt"]},
+                {"type": "text", "text": inputs["parser"].get_format_instructions()},
+                {"type": "image_url", "image_url": f"data:image/jpeg;base64,{inputs['image']}"}
+            ])
+        ]
+
+    def calculate_consumption(self, cb):
+        return {"total_cost": 0, "total_tokens": 0}
+
+
+class QueryEngineOllamaConversationalText(BaseQueryEngine):
+
+    def get_sample_message_content(self):
+        return """inputs = {"prompt": system_prompt + user_prompt}"""
+
+    def construct_message_content(self, inputs):
+        return [
+            HumanMessage(content=[
+                {"type": "text", "text": inputs["prompt"]}
+            ])
+        ]
+
+    def calculate_consumption(self, cb):
+        return {"total_cost": 0, "total_tokens": 0}
+
+
+class QueryEngineLlamacppConversationalText(BaseQueryEngine):
+
+    def get_sample_message_content(self):
+        return """inputs = {"system": system_prompt, "prompt": user_prompt}"""
+
+    def construct_message_content(self, inputs):
+        return [
+            {"role": "system", "content": inputs["system"]},
+            {"role": "user", "content": inputs["prompt"]}
+        ]
+
+    def calculate_consumption(self, cb):
+        return {"total_cost": 0, "total_tokens": cb.total_tokens}
+
+    def ask(self, inputs):
+        with get_openai_callback() as cb:
+            message_content = self.construct_message_content(inputs)
+            response = self.model.invoke(message_content)
+            logger.debug(f"Total Tokens: {cb.total_tokens}")
+            logger.debug(f"Prompt Tokens: {cb.prompt_tokens}")
+            logger.debug(f"Completion Tokens: {cb.completion_tokens}")
+            logger.debug(f"Total Cost (USD): ${round(cb.total_cost, 4)}")
+            consumption = self.calculate_consumption(cb)
+
+        answer = self.parse_response(response, inputs)
+        return answer, consumption

src/test.py

@@ -0,0 +1,10 @@
+import llms
+
+llm_factory = llms.LLMFactory()
+model_ollama = llm_factory.create_model(
+    provider="ollama",
+    conversational=True,
+    model_name="llama3.1:8b-instruct-fp16",
+)
+
+