Skip to main content
追踪 LangChain 有两种推荐方式：
  1. 将 LANGCHAIN_WANDB_TRACING 环境变量设置为 “true”。
  2. 使用上下文管理器 wandb_tracing_enabled() 追踪特定代码块。
注意：如果设置了环境变量，无论是否在上下文管理器中，所有代码都将被追踪。 
import os

from langchain_community.callbacks import wandb_tracing_enabled

os.environ["LANGCHAIN_WANDB_TRACING"] = "true"

# wandb 文档：使用环境变量配置 wandb
# https://docs.wandb.ai/guides/track/advanced/environment-variables
os.environ["WANDB_PROJECT"] = "langchain-tracing"

from langchain.agents import create_agent, load_tools
from langchain_openai import OpenAI

# 带追踪的 Agent 运行。确保 OPENAI_API_KEY 已正确设置才能运行此示例。

llm = OpenAI(temperature=0)
tools = load_tools(["llm-math"], llm=llm)

agent = create_agent(
    model=llm,
    tools=tools,
    verbose=True,
)

agent.invoke("What is 2 raised to .123243 power?")

# 现在，我们取消环境变量的设置，并使用上下文管理器。
if "LANGCHAIN_WANDB_TRACING" in os.environ:
    del os.environ["LANGCHAIN_WANDB_TRACING"]

# 使用上下文管理器启用追踪
with wandb_tracing_enabled():
    agent.invoke("What is 5 raised to .123243 power?")

agent.invoke("What is 2 raised to .123243 power?")

> Entering new AgentExecutor chain...
 I need to use a calculator to solve this.
Action: Calculator
Action Input: 5^.123243
Observation: Answer: 1.2193914912400514
Thought: I now know the final answer.
Final Answer: 1.2193914912400514

> Finished chain.


> Entering new AgentExecutor chain...
 I need to use a calculator to solve this.
Action: Calculator
Action Input: 2^.123243
Observation: Answer: 1.0891804557407723
Thought: I now know the final answer.
Final Answer: 1.0891804557407723

> Finished chain.

'1.0891804557407723'
在 GitHub 上编辑此页面或提交问题。
通过 MCP 将这些文档连接到 Claude、VSCode 等以获取实时答案。