Skip to main content

如何跟踪令牌使用情况

¥How to track token usage

Prerequisites

本指南假设你熟悉以下概念：

¥This guide assumes familiarity with the following concepts:

本注意本介绍了如何跟踪特定 LLM 调用的令牌使用情况。目前只有部分提供商实现了此功能，包括 OpenAI。

¥This notebook goes over how to track your token usage for specific LLM calls. This is only implemented by some providers, including OpenAI.

以下是通过回调跟踪单个 LLM 调用的令牌使用情况的示例：

¥Here's an example of tracking token usage for a single LLM call via a callback:

tip

请参阅 此部分包含有关安装集成包的一般说明。

¥See this section for general instructions on installing integration packages.

npm install @langchain/openai @langchain/core
import { OpenAI } from "@langchain/openai";

const llm = new OpenAI({
  model: "gpt-3.5-turbo-instruct",
  callbacks: [
    {
      handleLLMEnd(output) {
        console.log(JSON.stringify(output, null, 2));
      },
    },
  ],
});

await llm.invoke("Tell me a joke.");

/*
  {
    "generations": [
      [
        {
          "text": "\n\nWhy don't scientists trust atoms?\n\nBecause they make up everything.",
          "generationInfo": {
            "finishReason": "stop",
            "logprobs": null
          }
        }
      ]
    ],
    "llmOutput": {
      "tokenUsage": {
        "completionTokens": 14,
        "promptTokens": 5,
        "totalTokens": 19
      }
    }
  }
*/

API Reference:

如果将此模型传递给多次调用它的链或代理，它每次都会记录输出。

¥If this model is passed to a chain or agent that calls it multiple times, it will log an output each time.

后续步骤​

¥Next steps

现在你已经了解了如何获取受支持的 LLM 提供商的令牌使用情况。

¥You've now seen how to get token usage for supported LLM providers.

接下来，查看本节中的其他操作指南，例如 如何实现你自己的自定义 LLM。

¥Next, check out the other how-to guides in this section, like how to implement your own custom LLM.