{"id":1522,"date":"2024-05-19T07:37:55","date_gmt":"2024-05-19T07:37:55","guid":{"rendered":"https:\/\/www.nicekj.com\/?p=1522"},"modified":"2024-05-19T07:37:55","modified_gmt":"2024-05-19T07:37:55","slug":"langchainjs-shizhanxilierumenjieshao","status":"publish","type":"post","link":"https:\/\/www.nicekj.com\/langchainjs-shizhanxilierumenjieshao.html","title":{"rendered":"LangChain.js \u5b9e\u6218\u7cfb\u5217\uff1a\u5165\u95e8\u4ecb\u7ecd"},"content":{"rendered":"
\n

\ud83d\udcdd LangChain.js \u662f\u4e00\u4e2a\u5feb\u901f\u5f00\u53d1\u5927\u6a21\u578b\u5e94\u7528\u7684\u6846\u67b6\uff0c\u5b83\u63d0\u4f9b\u4e86\u4e00\u7cfb\u5217\u5f3a\u5927\u7684\u529f\u80fd\u548c\u5de5\u5177\uff0c\u4f7f\u5f97\u5f00\u53d1\u8005\u80fd\u591f\u66f4\u52a0\u9ad8\u6548\u5730\u6784\u5efa\u590d\u6742\u7684\u5e94\u7528\u7a0b\u5e8f\u3002LangChain.js \u5b9e\u6218\u7cfb\u5217\u6587\u7ae0\u5c06\u4ecb\u7ecd\u5728\u5b9e\u9645\u9879\u76ee\u4e2d\u4f7f\u7528 LangChain.js \u65f6\u7684\u4e00\u4e9b\u65b9\u6cd5\u548c\u6280\u5de7\u3002<\/p>\n<\/blockquote>\n

LangChain.js<\/a> \u662f\u4e00\u4e2a\u5feb\u901f\u6784\u5efa AI \u5e94\u7528\u7684\u5e93\uff0c\u5b83\u63d0\u4f9b\u4e86\u4e00\u7cfb\u5217\u7684\u5de5\u5177\uff0c\u53ef\u4ee5\u5e2e\u52a9\u4f60\u5feb\u901f\u6784\u5efa\u4e00\u4e2a AI \u5e94\u7528\u3002<\/p>\n

LangChain.js \u76ee\u524d\u8fd8\u5728\u5feb\u901f\u8fed\u4ee3\u4e2d\uff0c\u8fd9\u662f\u7531\u4e8e AI \u6280\u672f\u81ea\u8eab\u4e5f\u6b63\u5728\u5feb\u901f\u8fed\u4ee3\u4e2d\uff0c\u6240\u4ee5\u5f88\u591a\u529f\u80fd\u53ef\u80fd\u5f88\u5feb\u5c31\u88ab\u5e9f\u5f03\u6389\uff0c\u6bd4\u5982 generate()<\/code> \u65b9\u6cd5\u3002<\/p>\n

\u4f7f\u7528 LangChain.js \u7684\u597d\u5904\u6709\u633a\u591a\uff0c\u6bd4\u5982\uff1a<\/p>\n

    \n
  1. \u5c01\u88c5\u4e86\u5927\u91cf\u7684\u6a21\u578b\uff0c\u6bd4\u5982 OpenAI\u3001Azure OpenAI\u3001Claude\u3001\u6587\u5fc3\u4e00\u8a00\u7b49\u7b49\uff0c\u586b\u5165\u54cd\u5e94\u7684 API Key \u7b49\u53c2\u6570\u5373\u53ef\u8c03\u7528<\/li>\n
  2. \u63d0\u4f9b\u4e86\u5927\u91cf\u65b9\u4fbf\u7684\u65b9\u6cd5\uff0c\u6bd4\u5982\u94fe\u5f0f\u8c03\u7528\u3001\u5bf9\u8bdd\u7ba1\u7406\u3001\u56de\u94a9\u5b50\u7b49\u7b49<\/li>\n
  3. \u548c LangSmith \u7ed3\u5408\uff0c\u5bf9 AI \u5e94\u7528\u53ef\u4ee5\u5f88\u597d\u5730\u8fdb\u884c\u8c03\u8bd5\u5f00\u53d1<\/li>\n<\/ol>\n

    LangChain.js \u7684\u57fa\u672c\u4f7f\u7528<\/h2>\n

    \u8c03\u7528\u6a21\u578b<\/h3>\n

    LangChain.js \u65b0\u6539\u7248\u533a\u5206\u4e86\u4e24\u79cd\u8c03\u7528\u65b9\u5f0f\uff0c\u4e00\u79cd\u662fLLM<\/strong>\uff0c\u4e00\u79cd\u662fChatModel<\/strong>\uff0c\u4e0d\u8fc7\u8fd9\u4e24\u79cd\u8c03\u7528\u65b9\u5f0f\u672c\u8d28\u90fd\u4e00\u6837\uff0c\u6700\u7ec8\u90fd\u662f\u8c03\u7528\u6a21\u578b\uff0c\u4e00\u822c\u6211\u4eec\u4f7f\u7528\u540e\u8005\u3002<\/p>\n

    \u5b9e\u4f8b\u5316 ChatModel<\/code> \uff1a<\/p>\n

    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>import<\/span> { ChatOpenAI<\/span> } from<\/span> \"langchain\/chat_models\/openai\"<\/span>;<\/span>\n<\/span>\nconst<\/span> chatModel = new<\/span> ChatOpenAI<\/span>({<\/span>\n  openAIApiKey<\/span>: \"...\"<\/span>,<\/span>\n});<\/span>\n<\/code><\/pre>\n
    \u8fd9\u91cc openAIApiKey \u53ef\u4ee5\u5728\u5b9e\u4f8b\u5316\u7684\u65f6\u5019\u4f20\u5165\uff0c\u4e5f\u53ef\u4ee5\u653e\u7f6e\u5728\u73af\u5883\u53d8\u91cf OPENAI_API_KEY<\/code> \u4e2d\uff0c\u8fd9\u6837\u5c31\u4e0d\u7528\u6bcf\u6b21\u90fd\u4f20\u5165\u4e86\uff0cLangChain \u4f1a\u81ea\u52a8\u4ece process.env<\/code> \u8bfb\u53d6\u3002\u5982\u679c\u662f Azure OpenAI\uff0c\u90a3\u5bf9\u5e94\u7684\u5c31\u662f AZURE_OPENAI_API_KEY<\/code>\u3001AZURE_OPENAI_API_INSTANCE_NAME<\/code>\u3001AZURE_OPENAI_API_DEPLOYMENT_NAME<\/code> \u7b49\u7b49\u3002<\/p>\n
    \u63a5\u7740\u5c31\u53ef\u4ee5\u8c03\u7528\u6a21\u578b\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>import<\/span> { HumanMessage<\/span>, SystemMessage<\/span> } from<\/span> \"langchain\/chat_models\/messages\"<\/span>;<\/span>\n<\/span>\nconst<\/span> messages = [<\/span>\n  new<\/span> SystemMessage<\/span>(\"\u4f60\u662f\u4e00\u4f4d\u8bed\u8a00\u6a21\u578b\u4e13\u5bb6\"<\/span>),<\/span>\n  new<\/span> HumanMessage<\/span>(\"\u6a21\u578b\u6b63\u5219\u5316\u7684\u76ee\u7684\u662f\u4ec0\u4e48\uff1f\"<\/span>),<\/span>\n];<\/span>\n<\/code><\/pre>\n
    \u8fd9\u91cc\u7684 SystemMessage \u548c HumanMessage \u90fd\u662f LangChain.js \u63d0\u4f9b\u7684\u6d88\u606f\u7c7b\uff0c\u5206\u522b\u8868\u793a\u7cfb\u7edf\u6d88\u606f\u548c\u7528\u6237\u6d88\u606f\u3002\u7528\u6237\u6d88\u606f\u597d\u7406\u89e3\uff0c\u7cfb\u7edf\u6d88\u606f\u7684\u8bdd\u53ef\u4ee5\u770b\u4f5c\u662f\u9488\u5bf9 AI \u6a21\u578b\u7684\u4e00\u4e2a\u9ad8\u7ea7\u6307\u4ee4\uff08instruction\uff09\uff0c\u6bd4\u5982 SystemMessage(\"\u4f60\u662f\u4e00\u4f4d\u8bed\u8a00\u6a21\u578b\u4e13\u5bb6\")<\/code> \u5c31\u662f\u544a\u8bc9 AI \u6a21\u578b\uff0c\u4f60\u662f\u4e00\u4f4d\u8bed\u8a00\u6a21\u578b\u4e13\u5bb6\uff0c\u8fd9\u6837 AI \u6a21\u578b\u5c31\u4f1a\u4ee5\u8fd9\u4e2a\u8eab\u4efd\u6765\u56de\u7b54\u4f60\u7684\u95ee\u9898\uff0cSystemMessage<\/code> \u662f\u53ef\u9009\u7684\u3002<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>await<\/span> chatModel.invoke<\/span>(messages);<\/span>\n<\/code><\/pre>\n
    \u8fd9\u91cc\u7684 invoke()<\/code> \u65b9\u6cd5\u5c31\u662f\u8c03\u7528\u6a21\u578b\uff0c\u5b83\u4f1a\u8fd4\u56de\u4e00\u4e2a Promise<\/code>\uff0c\u8fd9\u4e2a Promise<\/code> \u7684\u7ed3\u679c\u5c31\u662f AI \u6a21\u578b\u7684\u56de\u590d\uff0c\u6bd4\u5982\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>AIMessage<\/span> { content<\/span>: 'The purpose of model regularization is to prevent overfitting in machine learning models. Overfitting occurs when a model becomes too complex and starts to fit the noise in the training data, leading to poor generalization on unseen data. Regularization techniques introduce additional constraints or penalties to the model'<\/span>s objective function<\/span>, discouraging it from<\/span> becoming overly complex and promoting simpler and more generalizable models. Regularization<\/span> helps to strike a balance between fitting the training data well and avoiding overfitting, leading to better performance on new<\/span>, unseen data.' }<\/span><\/span>\n<\/span>\n<\/code><\/pre>\n
    \u6d41\u5f0f\u4f20\u8f93<\/h3>\n
    \u6d41\u5f0f\u4f20\u8f93\u662f\u4e00\u4e2a\u57fa\u672c\u529f\u80fd\u4e86\uff0c\u4e00\u5f00\u59cb LangChain \u4ec5\u652f\u6301\u4f7f\u7528\u56de\u8c03\u51fd\u6570\u7684\u65b9\u5f0f\u6765\u5b9e\u73b0\uff0c\u6bd4\u5982\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> chat = new<\/span> ChatOpenAI<\/span>({<\/span>\n  streaming<\/span>: true<\/span>,<\/span>\n});<\/span>\n<\/span>\nconst<\/span> response = await<\/span> chat.call<\/span>([new<\/span> HumanMessage<\/span>(\"\u8bb2\u4e2a\u7b11\u8bdd\"<\/span>)], {<\/span>\n  callbacks<\/span>: [<\/span>\n    {<\/span>\n      handleLLMNewToken<\/span>(token: string<\/span><\/span>) {<\/span>\n        console<\/span>.log<\/span>({ token });<\/span>\n      },<\/span>\n    },<\/span>\n  ],<\/span>\n});<\/span>\n<\/code><\/pre>\n
    \u8fd9\u6837\u6bcf\u5f53\u6a21\u578b\u8fd4\u56de\u7684\u65f6\u5019\uff0c\u90fd\u4f1a\u89e6\u53d1 handleLLMNewToken<\/code> \u56de\u8c03\u51fd\u6570\uff0c\u65b0\u7248 LangChain.js \u66f4\u52a0\u7075\u6d3b\uff0c\u4f7f\u7528 .stream()<\/code> \u65b9\u6cd5\u53ef\u4ee5\u5b9e\u73b0\u540c\u6837\u7684\u529f\u80fd\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> stream = await<\/span> chat.stream<\/span>([new<\/span> HumanMessage<\/span>(\"\u8bb2\u4e2a\u7b11\u8bdd\"<\/span>)]);<\/span>\n<\/span>\nfor<\/span> await<\/span> (const<\/span> chunk of<\/span> stream) {<\/span>\n  console<\/span>.log<\/span>(chunk);<\/span>\n}<\/span>\n<\/code><\/pre>\n
    \u8fd9\u91cc\u7684 stream<\/code> \u662f\u4e00\u4e2a AsyncIterableIterator<\/code>\uff0c\u53ef\u4ee5\u4f7f\u7528 for await<\/code> \u6765\u904d\u5386\uff0c\u6bcf\u5f53\u6a21\u578b\u8fd4\u56de\u7684\u65f6\u5019\uff0c\u5c31\u4f1a\u89e6\u53d1 for await<\/code> \u4e2d\u7684\u4ee3\u7801\u3002<\/p>\n
    JSON Mode<\/h3>\n
    JSON Mode \u662f OpenAI \u65b0\u7248\u7684\u80fd\u529b\uff0c\u5b83\u53ef\u4ee5\u8ba9\u4f60\u66f4\u597d\u5730\u63a7\u5236 AI \u6a21\u578b\u7684\u8f93\u51fa\uff0c\u6bd4\u5982\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> jsonModeModel = new<\/span> ChatOpenAI<\/span>({<\/span>\n  modelName<\/span>: \"gpt-4-1106-preview\"<\/span>,<\/span>\n}).bind<\/span>({<\/span>\n  response_format<\/span>: {<\/span>\n    type<\/span>: \"json_object\"<\/span>,<\/span>\n  },<\/span>\n});<\/span>\n<\/code><\/pre>\n
    \u6ce8\u610f\uff0c\u76ee\u524d\u4ec5 gpt-4-1106-preview<\/code> \u6a21\u578b\u652f\u6301 JSON Mode\uff0c\u53e6\u5916\u8fd8\u6709\u4e00\u4e2a\u5f3a\u5236\u6027\u7684\u8981\u6c42\uff0c\u5c31\u662f SystemMessage<\/code> \u5fc5\u987b\u5305\u542b JSON<\/code> \u5b57\u773c\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> res = await<\/span> jsonModeModel.invoke<\/span>([<\/span>\n  [\"system\"<\/span>, \"Only return JSON\"<\/span>],<\/span>\n  [\"human\"<\/span>, \"Hi there!\"<\/span>],<\/span>\n]);<\/span>\n<\/code><\/pre>\n
    \u540e\u7eed GPT \u8fed\u4ee3 JSON Mode \u5e94\u8be5\u5c31\u4f1a\u53d8\u6210\u901a\u7528\u80fd\u529b\uff0c\u4e4b\u8bed SystemMessage<\/code> \u7684\u89c4\u5219\uff0c\u4e0d\u77e5\u9053\u540e\u7eed\u4f1a\u4e0d\u4f1a\u6539\u53d8\u3002<\/p>\n
    \u51fd\u6570\u8c03\u7528<\/h3>\n
    \u51fd\u6570\u8c03\u7528\uff08Function Calling\uff09\u662f OpenAI \u7684\u4e00\u4e2a\u91cd\u70b9\u80fd\u529b\uff0c\u4e5f\u5c31\u662f\u76ee\u524d AI \u5e94\u7528\u548c\u7a0b\u5e8f\u7684\u4e00\u4e2a\u91cd\u8981\u4ea4\u4e92\u534f\u8bae\u3002\u51fd\u6570\u8c03\u7528\u5176\u5b9e\u5f88\u7b80\u5355\uff0c\u5c31\u662f\u5148\u8ba9 AI \u53bb\u9009\u62e9\u8c03\u7528\u54ea\u4e2a\u51fd\u6570\uff0c\u7136\u540e\u5728\u7a0b\u5e8f\u4e2d\u8c03\u7528\u771f\u6b63\u7684\u51fd\u6570\u3002<\/p>\n
    \u6700\u5e38\u89c1\u7684\u573a\u666f\u5c31\u662f\u8054\u7f51\u56de\u7b54\uff0c\u4f60\u63d0\u4f9b\u4e86\u300c\u8054\u7f51\u641c\u7d22\u300d\u7684\u51fd\u6570\uff0c\u5f53\u7528\u6237\u63d0\u95ee\u300c\u4eca\u5929\u7684\u91cd\u70b9\u65b0\u95fb\u662f\u4ec0\u4e48\u300d\u7684\u65f6\u5019\uff0cAI \u4f1a\u5148\u8c03\u7528\u300c\u8054\u7f51\u641c\u7d22\u300d\u51fd\u6570\uff0c\u7136\u540e\u6839\u636e\u51fd\u6570\u6267\u884c\u5f97\u5230\u7684\u4fe1\u606f\uff0c\u6700\u7ec8\u518d\u56de\u7b54\u7528\u6237\u7684\u95ee\u9898\u3002<\/p>\n
    OpenAI \u4f7f\u7528 JSON Schema \u6765\u5b9a\u4e49\u51fd\u6570\u8c03\u7528\u7684\u534f\u8bae\uff0c\u6bd4\u5982\u5b9a\u4e49\u4e00\u4e2a\u63d0\u53d6\u5b57\u6bb5\u7684\u51fd\u6570\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> extractionFunctionSchema = {<\/span>\n  \/\/ \u5b9a\u4e49\u51fd\u6570\u7684\u540d\u5b57<\/span><\/span>\n  name<\/span>: \"extractor\"<\/span>,<\/span>\n  \/\/ \u5b9a\u4e49\u51fd\u6570\u7684\u63cf\u8ff0<\/span><\/span>\n  description<\/span>: \"Extracts fields from the input.\"<\/span>,<\/span>\n  \/\/ \u5b9a\u4e49\u51fd\u6570\u7684\u5165\u53c2\u6709\u54ea\u4e9b<\/span><\/span>\n  parameters<\/span>: {<\/span>\n    type<\/span>: \"object\"<\/span>,<\/span>\n    properties<\/span>: {<\/span>\n      tone<\/span>: {<\/span>\n        type<\/span>: \"string\"<\/span>,<\/span>\n        enum<\/span>: [\"positive\"<\/span>, \"negative\"<\/span>],<\/span>\n        description<\/span>: \"The overall tone of the input\"<\/span>,<\/span>\n      },<\/span>\n      word_count<\/span>: {<\/span>\n        type<\/span>: \"number\"<\/span>,<\/span>\n        description<\/span>: \"The number of words in the input\"<\/span>,<\/span>\n      },<\/span>\n      chat_response<\/span>: {<\/span>\n        type<\/span>: \"string\"<\/span>,<\/span>\n        description<\/span>: \"A response to the human's input\"<\/span>,<\/span>\n      },<\/span>\n    },<\/span>\n    required<\/span>: [\"tone\"<\/span>, \"word_count\"<\/span>, \"chat_response\"<\/span>],<\/span>\n  },<\/span>\n};<\/span>\n<\/code><\/pre>\n
    \u4e5f\u53ef\u4ee5\u4f7f\u7528 zod<\/code> \u8fd9\u4e2a\u5e93\uff0c\u5199\u8d77\u6765\u66f4\u65b9\u4fbf\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>import<\/span> { z } from<\/span> \"zod\"<\/span>;<\/span>\nimport<\/span> { zodToJsonSchema } from<\/span> \"zod-to-json-schema\"<\/span>;<\/span>\n<\/span>\nconst<\/span> extractionFunctionSchema = {<\/span>\n  name<\/span>: \"extractor\"<\/span>,<\/span>\n  description<\/span>: \"Extracts fields from the input.\"<\/span>,<\/span>\n  parameters<\/span>: zodToJsonSchema<\/span>(<\/span>\n    z.object<\/span>({<\/span>\n      tone<\/span>: z<\/span>\n        .enum<\/span>([\"positive\"<\/span>, \"negative\"<\/span>])<\/span>\n        .describe<\/span>(\"The overall tone of the input\"<\/span>),<\/span>\n      entity<\/span>: z.string<\/span>().describe<\/span>(\"The entity mentioned in the input\"<\/span>),<\/span>\n      word_count<\/span>: z.number<\/span>().describe<\/span>(\"The number of words in the input\"<\/span>),<\/span>\n      chat_response<\/span>: z.string<\/span>().describe<\/span>(\"A response to the human's input\"<\/span>),<\/span>\n      final_punctuation<\/span>: z<\/span>\n        .optional<\/span>(z.string<\/span>())<\/span>\n        .describe<\/span>(\"The final punctuation mark in the input, if any.\"<\/span>),<\/span>\n    })<\/span>\n  ),<\/span>\n};<\/span>\n<\/code><\/pre>\n
    \u8c03\u7528\u51fd\u6570\uff1a<\/p>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> model = new<\/span> ChatOpenAI<\/span>({<\/span>\n  modelName<\/span>: \"gpt-4\"<\/span>,<\/span>\n}).bind<\/span>({<\/span>\n  functions<\/span>: [extractionFunctionSchema],<\/span>\n  function_call<\/span>: { name<\/span>: \"extractor\"<\/span> },<\/span>\n});<\/span>\n<\/code><\/pre>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>const<\/span> result = await<\/span> model.invoke<\/span>([new<\/span> HumanMessage<\/span>(\"What a beautiful day!\"<\/span>)]);<\/span>\n<\/code><\/pre>\n
    <\/div>
    typescript<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>console<\/span>.log<\/span>(result);<\/span>\n\/*<\/span><\/span>\nAIMessage {<\/span>\n  \/\/...<\/span>\n  additional_kwargs: {<\/span>\n    function_call: {<\/span>\n      name: 'extractor',<\/span>\n      arguments: '{\\n' +<\/span>\n        '\"tone\": \"positive\",\\n' +<\/span>\n        '\"entity\": \"day\",\\n' +<\/span>\n        '\"word_count\": 4,\\n' +<\/span>\n        `\"chat_response\": \"I'm glad you're enjoying the day!\",\\n` +<\/span>\n        '\"final_punctuation\": \"!\"\\n' +<\/span>\n        '}'<\/span>\n    }<\/span>\n  }<\/span>\n}<\/span>\n*\/<\/span>\n<\/code><\/pre>\n
    \u6700\u540e<\/h2>\n
    \u63a8\u8350\u4e00\u4e9b\u597d\u7528\u7684\u8d44\u6e90<\/p>\n
    \ud83d\udc49   StarFlow.tech <\/a> \uff0c\u4e00\u4e2a\u96c6\u804a\u5929\u3001\u5de5\u4f5c\u6d41\u548c\u77e5\u8bc6\u5e93\u7684 AI \u5e73\u53f0\u3002\u5728\u8fd9\u91cc\uff0c\u4f60\u53ef\u4ee5\u514d\u8d39\u4f7f\u7528 ChatGPT3.5 \u548c 3.5 16K\uff0c\u8fd8\u6709 GPT-4 Vision\u3001DELL\u00b7E3\u3001Midjourney \u7b49\u591a\u79cd\u6a21\u578b\u53ef\u4f9b\u9009\u62e9\u3002\u8fd9\u4e2a\u5e73\u53f0\u5c31\u50cf\u4e00\u4e2a\u5c0f\u578b\u5de5\u4f5c\u5ba4\uff0c\u52a9\u529b\u4e2a\u4eba\u6548\u7387 Max\uff01<\/p>\n
    \ud83d\udc49 OpenAI \u5b98\u65b9\u63d0\u793a\u8bcd\u6307\u5357<\/a> \uff0c\u4e13\u95e8\u9762\u5411\u4e2d\u6587\u7684\u63d0\u793a\u8bcd\u5de5\u7a0b\u6307\u5357\uff0c\u8be5\u6559\u7a0b\u662f OpenAI \u5b98\u65b9\u51fa\u7248\uff0c\u4e3b\u8981\u5305\u62ec\u4e86\u516d\u5927\u7b56\u7565\uff0c\u8f7b\u677e\u5b66\u4e60\u63d0\u793a\u8bcd\u6280\u5de7\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
    LangChain.js \u662f\u4e00\u4e2a\u5feb\u901f\u6784\u5efa AI \u5e94\u7528\u7684\u5e93\uff0c\u5b83\u63d0\u4f9b\u4e86\u4e00\u7cfb\u5217\u7684\u5de5\u5177\uff0c\u53ef\u4ee5\u5e2e\u52a9\u4f60\u5feb\u901f\u6784\u5efa\u4e00\u4e2a AI \u5e94\u7528\u3002 LangChain.js \u76ee\u524d\u8fd8\u5728\u5feb\u901f\u8fed\u4ee3\u4e2d\uff0c\u8fd9\u662f\u7531\u4e8e AI \u6280\u672f\u81ea\u8eab\u4e5f\u6b63\u5728\u5feb\u901f\u8fed<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_title":"","rank_math_description":"","rank_math_focus_keyword":"","views":"1","footnotes":""},"categories":[3],"tags":[136,126,127,128,129],"collection":[],"class_list":["post-1522","post","type-post","status-publish","format-standard","hentry","category-fenlei2","tag-136","tag-gpt","tag-ai","tag-128","tag-129"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1522","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/comments?post=1522"}],"version-history":[{"count":0,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1522\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media?parent=1522"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/categories?post=1522"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/tags?post=1522"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/collection?post=1522"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}