Migrating LLMChain to v0.3

[santurini]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

def create_llm_chain(model, prompt_path):
    prompt_template = read_txt(prompt_path)
    prompt = PromptTemplate(
        template=prompt_template,
        input_variables=["context", "question"]
    )
    llm = ChatOpenAI(model=model)
    memory = ConversationBufferMemory(memory_key="history", input_key="question")
    llm_chain = LLMChain(
        llm=llm,
        prompt=prompt,
        memory=memory
    )

    return llm_chain

Description

How can I migrate this LLMChain to v0.3 without losing functionalities?

System Info

System Information

OS: Windows
OS Version: 10.0.22631
Python Version: 3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)]

Package Information

langchain_core: 0.3.2
langchain: 0.2.5
langchain_community: 0.2.5
langsmith: 0.1.125
langchain_cli: 0.0.31
langchain_huggingface: 0.0.3
langchain_openai: 0.1.8
langchain_text_splitters: 0.2.1
langserve: 0.3.0

Optional packages not installed

langgraph

Other Dependencies

aiohttp: 3.10.5
async-timeout: 4.0.3
beautifulsoup4: Installed. No version info available.
dataclasses-json: 0.6.7
fastapi: 0.112.1
gitpython: 3.1.43
gritql: 0.1.4
httpx: 0.27.0
huggingface-hub: 0.24.6
jsonpatch: 1.33
langserve[all]: Installed. No version info available.
lxml: Installed. No version info available.
numpy: 1.26.4
openai: 1.42.0
orjson: 3.10.7
packaging: 24.1
pydantic: 2.8.2
PyYAML: 6.0.2
requests: 2.32.3
sentence-transformers: 3.0.1
SQLAlchemy: 2.0.32
sse-starlette: 1.8.2
tenacity: 8.5.0
tiktoken: 0.7.0
tokenizers: 0.19.1
tomlkit: 0.12.5
transformers: 4.44.1
typer[all]: Installed. No version info available.
typing-extensions: 4.12.2
uvicorn: 0.23.2

[dosubot[bot]]

To migrate your LLMChain to LangChain v0.3 without losing functionalities, follow these steps:

1. Upgrade your LangChain packages to the recommended versions for v0.3:

   pip install langchain>=0.3,<0.4
   pip install langchain-core>=0.3,<0.4

2. Update Pydantic imports from langchain_core.pydantic_v1 or langchain.pydantic_v1 to pydantic:

   from pydantic import BaseModel

3. Ensure Pydantic 2 compatibility in your code. If you have any Pydantic 1 specific code, update it to Pydantic 2. For example, replace validator with field_validator:

   from pydantic import Field, field_validator
   
   class CustomTool(BaseTool):
       x: int = Field(default=1)
   
       def _run(*args, **kwargs):
           return "hello"
   
       @field_validator('x')
       @classmethod
       def validate_x(cls, x: int) -> int:
           return 1

4. Rebuild models if necessary using model_rebuild():

   from langchain_core.output_parsers import BaseOutputParser
   
   class FooParser(BaseOutputParser):
       ...
   
   FooParser.model_rebuild()

5. Use the langchain-cli tool to automatically update deprecated imports:

   pip install -U langchain-cli
   langchain-cli migrate [path to code]
   langchain-cli migrate [path to code] # Run a second time to apply more import replacements

6. Migrate to LCEL and LangGraph: LangChain has deprecated many of the original "Chain" classes in favor of the more flexible and powerful frameworks of LCEL (LangChain Expression Language) and LangGraph. These frameworks provide a unified interface, composition primitives, and built-in persistence, among other features. You can reimplement your LLMChain using these new abstractions for better performance and maintainability. Refer to the LLMChain migration guide for detailed steps.

Here is an example of how you can migrate your LLMChain using LangGraph:

import math
from typing import Annotated, Sequence

import numexpr
from langchain_core.messages import BaseMessage
from langchain_core.runnables import RunnableConfig
from langchain_core.tools import tool
from langchain_openai import ChatOpenAI
from langgraph.graph import END, StateGraph
from langgraph.graph.message import add_messages
from langgraph.prebuilt.tool_node import ToolNode
from typing_extensions import TypedDict

@tool
def calculator(expression: str) -> str:
    local_dict = {"pi": math.pi, "e": math.e}
    return str(
        numexpr.evaluate(
            expression.strip(),
            global_dict={},  # restrict access to globals
            local_dict=local_dict,  # add common mathematical functions
        )
    )

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0)
tools = [calculator]
llm_with_tools = llm.bind_tools(tools, tool_choice="any")

class ChainState(TypedDict):
    messages: Annotated[Sequence[BaseMessage], add_messages]

async def acall_chain(state: ChainState, config: RunnableConfig):
    last_message = state["messages"][-1]
    response = await llm_with_tools.ainvoke(state["messages"], config)
    return {"messages": [response]}

async def acall_model(state: ChainState, config: RunnableConfig):
    response = await llm.ainvoke(state["messages"], config)
    return {"messages": [response]}

graph_builder = StateGraph(ChainState)
graph_builder.add_node("call_tool", acall_chain)
graph_builder.add_node("execute_tool", ToolNode(tools))
graph_builder.add_node("call_model", acall_model)
graph_builder.set_entry_point("call_tool")
graph_builder.add_edge("call_tool", "execute_tool")
graph_builder.add_edge("execute_tool", "call_model")
graph_builder.add_edge("call_model", END)
chain = graph_builder.compile()

By following these steps, you can migrate your LLMChain to LangChain v0.3 while maintaining all functionalities and taking advantage of the new features and improvements ^[1][2][3].

To continue talking to Dosu, mention @dosu.

[eyurtsev]

Use langgraph if you need memory: https://python.langchain.com/docs/tutorials/agents/#adding-in-memory

You can also follow this tutorial which just langchain without any langgraph: https://python.langchain.com/docs/tutorials/chatbot/