from langchain.agents import load_tools from langchain.agents import initialize_agent from langchain.agents import AgentType from langchain.llms import OpenAI # First, let's load the language model we're going to use to control the agent. llm = OpenAI(temperature=0) # Next, let's load some tools to use. Note that the `llm-math` tool uses an LLM, so we need to pass that in. tools = load_tools(["serpapi", "llm-math"], llm=llm) # Finally, let's initialize an agent with the tools, the language model, and the type of agent we want to use. agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True) # Now let's test it out! agent.run("昨天北京气温如何?")