LangChain

To integrate our tracing features with LangChain, you simply need to use a callback handler, BaserunCallbackHandler, which you can pass when creating a client object or a chain. Currently, only our Python SDK supports integration with LangChain.

Tracing a client

If you use the BaserunCallbackHandler you do not need to use @baserun.trace- the handler does it for you.

python

from baserun.instrumentation.langchain import BaserunCallbackHandler
from langchain.chat_models import ChatOpenAI
from langchain_core.messages import HumanMessage

def use_langchain(question="What is the capital of the US?") -> str:
    chat = ChatOpenAI(callbacks=[BaserunCallbackHandler()])
    messages = [HumanMessage(content=question)]
    response = chat.invoke(messages)
    return response.content

Tracing a chain

To trace a chain, pass in the BaserunCallbackHandler to the callbacks parameter of LLMChain.

python

from langchain.chat_models import ChatOpenAI
from langchain_core.prompts import PromptTemplate
from langchain.chains import LLMChain
from baserun.instrumentation.langchain import BaserunCallbackHandler

def use_langchain_chain(question="What is the capital of {location}?") -> str:
    chat = ChatOpenAI()

    chain = LLMChain(
        llm=chat,
        prompt=PromptTemplate(template=question, input_variables=["location"]),
        callbacks=[BaserunCallbackHandler()],
    )
    response = chain.run(location="California")
    return response

Tracing an agent

To trace an agent, pass in the BaserunCallbackHandler to the callback_manager parameter of initialize_agent.

python

from langchain.chat_models import ChatOpenAI
from langchain_core.prompts import PromptTemplate
from langchain.agents import initialize_agent
from langchain.agents import AgentType
from langchain.agents import load_tools
from langchain_core.callbacks import BaseCallbackManager
from baserun.instrumentation.langchain import BaserunCallbackHandler

def use_langchain_agent_tools(question="Using Wikipedia, look up the population of {location} as of 2023.") -> str:
    # Note: To run this, you must `pip install wikipedia langchain`
    chat = ChatOpenAI()
    tools = load_tools(["wikipedia"])
    prompt_template = PromptTemplate(template=question, input_variables=["location"])

    agent = initialize_agent(
        tools,
        chat,
        agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
        callback_manager=BaseCallbackManager(handlers=[BaserunCallbackHandler()]),
    )
    response = agent.run(prompt_template.format_prompt(location="California"))
    return response

Introduction

Prompt playground

Python SDK 2.0

Get started with SDK

Monitoring

Prompt templates

Evaluation

Testing

Datasets

Fine-tune

Integrations

LangChain

LangChain

Tracing a client

Tracing a chain

Tracing an agent

​LangChain

​Tracing a client

​Tracing a chain

​Tracing an agent

LangChain

Tracing a client

Tracing a chain

Tracing an agent