函数调用(Function calling)详解 / ai #31

函数调用(Function calling)可以极大地增强大语言模型的功能，可以增强推理效果或进行其他外部操作，包括信息检索、数据库操作、知识图谱搜索与推理、操作系统、触发外部操作等工具调用场景。在设计各类bot或agent时非常有用。

函数调用图示

如上所示，在第二步中，可以提供特定的接口函数（也就是外部API），额外增加特定的数据以增强大语言模型的能力。在特定的工作流中，比如查询天气，查询车次等，大语言模型都能够办到啰！

下面我使用的是OpenAI中的示例，进行测试后，得到的结果和代码，大家可以参考。

import OpenAI from "openai"
import dotEnv from "dotenv"

dotEnv.config()
const apiKey = process.env.API_KEY

const openai = new OpenAI({
  apiKey: apiKey 
})

// Example dummy function hard coded to return the same weather
// In production, this could be your backend API or an external API
function getCurrentWeather(location, unit = "fahrenheit") {
  if (location.toLowerCase().includes("tokyo")) {
    return JSON.stringify({ location: "Tokyo", temperature: "10", unit: "celsius" });
  } else if (location.toLowerCase().includes("san francisco")) {
    return JSON.stringify({ location: "San Francisco", temperature: "72", unit: "fahrenheit" });
  } else if (location.toLowerCase().includes("paris")) {
    return JSON.stringify({ location: "Paris", temperature: "22", unit: "fahrenheit" });
  } else {
    return JSON.stringify({ location, temperature: "unknown" });
  }
}

async function runConversation() {
  // Step 1: send the conversation and available functions to the model
  const messages = [
    { role: "user", content: "What's the weather like in San Francisco, Tokyo, and Paris?" },
  ];

  const tools = [
    {
      type: "function",
      function: {
        name: "get_current_weather",
        description: "Get the current weather in a given location",
        parameters: {
          type: "object",
          properties: {
            location: {
              type: "string",
              description: "The city and state, e.g. San Francisco, CA",
            },
            unit: { type: "string", enum: ["celsius", "fahrenheit"] },
          },
          required: ["location"],
        },
      },
    },
  ];


  const response = await openai.chat.completions.create({
    model: "gpt-3.5-turbo-0125",
    messages: messages,
    tools: tools,
    tool_choice: "auto", // auto is default, but we'll be explicit
  });
  const responseMessage = response.choices[0].message;
  console.log(111, "response", response)
  /*
  {
    id: 'chatcmpl-90oyNdEDQSxsvpMpwxZ3KeYShovtH',
    object: 'chat.completion',
    created: 1709982967,
    model: 'gpt-3.5-turbo-0125',
    choices: [
      {
        index: 0,
        message: [Object],
        logprobs: null,
        finish_reason: 'tool_calls'
      }
    ],
    usage: { prompt_tokens: 88, completion_tokens: 77, total_tokens: 165 },
    system_fingerprint: 'fp_4f0b692a78'
  }
  */

  console.log(123, "responseMessage", responseMessage)
  /* promptX
  {
  role: 'assistant',
  content: null,
  tool_calls: [
    {
      id: 'call_0QMDlATqYjBjgdySQVCL0PI4',
      type: 'function',
      function: [Object]
    },
    {
      id: 'call_e8ceipN3i2OUiF3Z38qNEEs5',
      type: 'function',
      function: [Object]
    },
    {
      id: 'call_cHlSgOIcsOXwpiOpzHIev6vF',
      type: 'function',
      function: [Object]
    }
  ]
  }
  */

  //console.log(396, "function", responseMessage.tool_calls[0].function)
  /*
  {
    name: 'get_current_weather',
    arguments: '{"location": "San Francisco", "unit": "celsius"}'
  }
  */

  // Step 2: check if the model wanted to call a function
  const toolCalls = responseMessage.tool_calls
  if (responseMessage.tool_calls) {
    // call the function
    // Note: the JSON response may not always be valid; be sure to handle errors
    const availableFunctions = {
      get_current_weather: getCurrentWeather,
    }; // only one function in this example, but you can have multiple
    messages.push(responseMessage); // extend conversation with assistant's reply
    console.log(225,"messages", messages)
    /*
    [{
      role: 'user',
      content: "What's the weather like in San Francisco, Tokyo, and Paris?"
    },
    {
      role: 'assistant',
      content: null,
      tool_calls: [ [Object], [Object], [Object] ]
    }]
    */

    for (const toolCall of toolCalls) {
      const functionName = toolCall.function.name;
      const functionToCall = availableFunctions[functionName];
      const functionArgs = JSON.parse(toolCall.function.arguments);
      const functionResponse = functionToCall(
        functionArgs.location,
        functionArgs.unit
      );
      messages.push({
        tool_call_id: toolCall.id,
        role: "tool",
        name: functionName,
        content: functionResponse,
      }); // extend conversation with function response
    }
    console.log(365,"messages2", messages)
  /*
  [{
    role: 'user',
    content: "What's the weather like in San Francisco, Tokyo, and Paris?"
  },
  {
    role: 'assistant',
    content: null,
    tool_calls: [ [Object], [Object], [Object] ]
  },
  {
    tool_call_id: 'call_0QMDlATqYjBjgdySQVCL0PI4',
    role: 'tool',
    name: 'get_current_weather',
    content: '{"location":"San Francisco","temperature":"72","unit":"fahrenheit"}'
  },
  {
    tool_call_id: 'call_e8ceipN3i2OUiF3Z38qNEEs5',
    role: 'tool',
    name: 'get_current_weather',
    content: '{"location":"Tokyo","temperature":"10","unit":"celsius"}'
  },
  {
    tool_call_id: 'call_cHlSgOIcsOXwpiOpzHIev6vF',
    role: 'tool',
    name: 'get_current_weather',
    content: '{"location":"Paris","temperature":"22","unit":"fahrenheit"}'
  }]
  */
    
    //Step 3: get a new response from the model where it can see the function response
    const secondResponse = await openai.chat.completions.create({
      model: "gpt-3.5-turbo-0125",
      messages: messages,
    }); 
    console.log(569, secondResponse)
    return secondResponse.choices;
  }
}


runConversation().then(console.log).catch(console.error);

经过三步之后会得到最终的结果，目前来看效果还不错！其实函数调用(Function calling)和外挂向量数据库有类似的地方，都是要给大语言模型提供额外的信息以得到更为准确的结果。 如果你正在开发bot或agent，那就用起来吧，它会使大语言模型的能力得到拓展！