支持结构化输出的智能体¶

当您运行智能体或多智能体框架时，可能需要它以特定格式输出结果。本笔记本将通过一个简单示例，展示如何为FunctionAgent实现这一功能！🦙🚀

首先安装所需的依赖项

In [ ]:

! pip install llama-index

! pip install llama-index

In [ ]:

from getpass import getpass
import os

os.environ["OPENAI_API_KEY"] = getpass()

from getpass import getpass import os os.environ["OPENAI_API_KEY"] = getpass()

现在我们来定义结构化输出格式

In [ ]:

from pydantic import BaseModel, Field


class MathResult(BaseModel):
    operation: str = Field(description="The operation that has been performed")
    result: int = Field(description="Result of the operation")

from pydantic import BaseModel, Field class MathResult(BaseModel): operation: str = Field(description="The operation that has been performed") result: int = Field(description="Result of the operation")

以及一个非常简单的计算器代理

In [ ]:

from llama_index.llms.openai import OpenAI
from llama_index.core.agent.workflow import FunctionAgent

llm = OpenAI(model="gpt-4.1")


def add(x: int, y: int):
    """Add two numbers"""
    return x + y


def multiply(x: int, y: int):
    """Multiply two numbers"""
    return x * y


agent = FunctionAgent(
    llm=llm,
    output_cls=MathResult,
    tools=[add, multiply],
    system_prompt="You are a calculator agent that can add or multiply two numbers by calling tools",
    name="calculator",
)

from llama_index.llms.openai import OpenAI from llama_index.core.agent.workflow import FunctionAgent llm = OpenAI(model="gpt-4.1") def add(x: int, y: int): """Add two numbers""" return x + y def multiply(x: int, y: int): """Multiply two numbers""" return x * y agent = FunctionAgent( llm=llm, output_cls=MathResult, tools=[add, multiply], system_prompt="You are a calculator agent that can add or multiply two numbers by calling tools", name="calculator", )

现在让我们运行代理程序

In [ ]:

response = await agent.run("What is the result of 10 multiplied by 4?")

response = await agent.run("What is the result of 10 multiplied by 4?")

最后，我们可以得到结构化输出

In [ ]:

# print the structured output as a plain dictionary
print(response.structured_response)
# print the structured output as a Pydantic model
print(response.get_pydantic_model(MathResult))

# print the structured output as a plain dictionary print(response.structured_response) # print the structured output as a Pydantic model print(response.get_pydantic_model(MathResult))