{"id":19705,"date":"2025-11-28T15:16:58","date_gmt":"2025-11-28T07:16:58","guid":{"rendered":"https:\/\/92it.top\/?p=19705"},"modified":"2025-11-28T15:16:58","modified_gmt":"2025-11-28T07:16:58","slug":"%e4%bb%80%e4%b9%88%e6%98%aflangchain%ef%bc%9f%e4%bb%8e%e9%9b%b6%e5%bc%80%e5%a7%8b%e5%85%a5%e9%97%a8-langchain","status":"publish","type":"post","link":"https:\/\/92it.top\/?p=19705","title":{"rendered":"\u4ec0\u4e48\u662fLangChain\uff1f\u4ece\u96f6\u5f00\u59cb\u5165\u95e8 LangChain"},"content":{"rendered":"\n<p>\u8f6c\u8f7d\uff1a<a href=\"https:\/\/blog.csdn.net\/2401_85375151\/article\/details\/147255146\">https:\/\/blog.csdn.net\/2401_85375151\/article\/details\/147255146<\/a><\/p>\n\n\n\n<p>\u672c\u7bc7\u6587\u7ae0\u4e2d\u5c06\u548c\u5927\u5bb6\u8bb2\u8ff0\u4ec0\u4e48\u662f LangChain \uff0c\u4ee5\u53ca LangChain \u89e3\u51b3\u4e86\u73b0\u5728\u5927\u6a21\u578b\u53d1\u5c55\u7684\u54ea\u4e9b\u95ee\u9898\uff0c\u7136\u540e\u4f1a\u8bb2\u89e3LangChain \u4e2d\u57fa\u7840\u7684\u6982\u5ff5\u548c\u7ec4\u4ef6\u3002\u5728\u6b64\u57fa\u7840\u4e0a\uff0c\u6211\u4eec\u4f1a\u6df1\u5165\u7684\u5206\u6790\u4e00\u4e0b LangChain \u5177\u4f53\u6709\u54ea\u4e00\u4e9b\u4f18\u52bf\uff0c\u6700\u540e\uff0c\u5c06\u5411\u5927\u5bb6\u4ecb\u7ecd\u4e00\u4e9b\u5f00\u6e90\u7684\u57fa\u4e8e LangChain \u7684\u9879\u76ee\uff0c\u6765\u611f\u53d7\u8fd9\u4e2a\u6846\u67b6\u5728\u5b9e\u9645\u4ea7\u54c1\u4e2d\u662f\u5982\u4f55\u4f7f\u7528\u548c\u843d\u5730\u7684\u3002<\/p>\n\n\n\n<p><strong>\u4e00\u3001LangChain \u662f\u4ec0\u4e48\ud83d\udd16<\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>LangChain \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u57fa\u4e8e LLM \u7684\u4e0a\u5c42\u5e94\u7528\u5f00\u53d1\u6846\u67b6\uff0cLangChain \u63d0\u4f9b\u4e86\u4e00\u7cfb\u5217\u7684\u5de5\u5177\u548c\u63a5\u53e3\uff0c\u8ba9\u5f00\u53d1\u8005\u53ef\u4ee5\u8f7b\u677e\u5730\u6784\u5efa\u548c\u90e8\u7f72\u57fa\u4e8e LLM \u7684\u5e94\u7528 \u3002LangChain \u56f4\u7ed5\u5c06\u4e0d\u540c\u7ec4\u4ef6\u201c\u94fe\u63a5\u201d\u5728\u4e00\u8d77\u7684\u6838\u5fc3\u6982\u5ff5\u6784\u5efa\uff0c\u7b80\u5316\u4e86\u4e0e GPT-3.5\u3001GPT-4 \u7b49 LLM \u5408\u4f5c\u7684\u8fc7\u7a0b\uff0c\u4f7f\u5f97\u6211\u4eec\u53ef\u4ee5\u8f7b\u677e\u521b\u5efa\u5b9a\u5236\u7684\u9ad8\u7ea7\u7528\u4f8b\u3002<\/p>\n\n\n\n<p>LangChain \u5df2\u7ecf\u6210\u4e3a\u5927\u6a21\u578b\u5e94\u7528\u5f00\u53d1\u7684\u6700\u4e3b\u6d41\u6846\u67b6\uff08\u4e4b\u4e00\uff09,23 \u5e74 1 \u6708\u7684\u4f17\u591a AI Hackathon \u51b3\u8d5b\u9879\u76ee\u4f7f\u7528 LangChain\u30022023 \u5e74\u878d\u8d44 3000w+ \u7f8e\u5200\uff08\u7ea2\u6749\uff09\u3002<\/p>\n\n\n\n<p>\u76ee\u524d\uff0c LangChain \u652f\u6301 Python \u548c TypeScript \u4e24\u79cd\u8bed\u8a00\u3002\u5982\u679c\u5927\u5bb6\u60f3\u4f53\u9a8c LangChain \uff0c\u5efa\u8bae\u4f7f\u7528 python \u8bed\u8a00\uff0c\u7b80\u5355\uff0c\u6613\u4e0a\u624b\u3002<\/p>\n\n\n\n<p>LangChain \u7684\u5b98\u7f51\u662f\u00a0<a href=\"https:\/\/link.juejin.cn\/?target=https%3A%2F%2Fpython.langchain.com%2Fdocs%2Fget_started%2Fintroduction\">LangChain\u5b98\u7f51<\/a><\/p>\n\n\n\n<p>\u5728\u8fd9\u4e0a\u9762\u53ef\u4ee5\u627e\u5230\u6240\u6709\u7684\u4f7f\u7528\u6848\u4f8b\u548c\u6559\u7a0b\u4fe1\u606f\u3002<\/p>\n\n\n\n<p><strong>\u4e8c\u3001LangChain \u89e3\u51b3\u4e86\u4ec0\u4e48\u95ee\u9898 <strong>\ud83d\udd16<\/strong><\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u901a\u8fc7\u4e0a\u9762\u7684\u6982\u5ff5\uff0c\u6211\u4eec\u53ef\u4ee5\u770b\u5230 LangChain \u5b9e\u9645\u4e0a\u662f\u57fa\u4e8e\u5927\u8bed\u8a00\u6a21\u578b\u4e0a\u5c42\u7684\u4e00\u4e2a\u5e94\u7528\u6846\u67b6\uff0c\u90a3\u4e48 LangChain \u5177\u4f53\u89e3\u51b3\u4e86\u5927\u6a21\u578b\u65f6\u4ee3\u7684\u54ea\u4e9b\u95ee\u9898\u624d\u8ba9\u4ed6\u8131\u9896\u800c\u51fa\u5462\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u4e3b\u8981\u6709\u4ee5\u4e0b\u51e0\u4e2a\u65b9\u9762\uff1a<\/p>\n\n\n\n<p><strong>1.\u6a21\u578b\u63a5\u53e3\u7684\u7edf\u4e00<\/strong><\/p>\n\n\n\n<p>\u73b0\u5728\u7684\u5927\u6a21\u578b\u9664\u4e86\u5927\u5bb6\u719f\u77e5\u7684 ChatGPT\uff0c\u8fd8\u6709 Meta \u5f00\u6e90\u7684 LLaMA\uff0c\u6e05\u534e\u5927\u5b66\u7684 GLM \u7b49\uff0c\u8fd9\u4e9b\u6a21\u578b\u7684\u4f7f\u7528\u65b9\u6cd5\u5305\u62ec api \u548c\u63a8\u7406\u65b9\u5f0f\u90fd\u76f8\u5dee\u751a\u8fdc\uff0c\u5982\u679c\u4f60\u60f3\u4ece\u4f7f\u7528 ChatGPT \u5207\u6362\u5230\u8c03\u7528 LLaMA\uff0c\u9700\u8981\u82b1\u8d39\u4e0d\u5c11\u7684\u7cbe\u529b\u53bb\u5f00\u53d1\u524d\u7f6e\u7684\u6a21\u578b\u4f7f\u7528\u6a21\u5757\uff0c\u4f1a\u6709\u5927\u91cf\u91cd\u590d\u7e41\u7410\u7684\u5de5\u4f5c\u3002\u800c LangChain \u5bf9\u597d\u591a\u5e38\u89c1\u7684 API \u548c\u5927\u6a21\u578b\u505a\u4e86\u5c01\u88c5\uff0c\u53ef\u4ee5\u76f4\u63a5\u62ff\u6765\u5c31\u7528\uff0c\u8282\u7701\u4e86\u5927\u91cf\u7684\u65f6\u95f4\u3002<\/p>\n\n\n\n<p><strong>2.\u6253\u7834\u4e86 LLM \u63d0\u793a\u8bcd\u548c\u8fd4\u56de\u5185\u5bb9 token \u9650\u5236\uff0c\u4e3a\u6700\u65b0\u77e5\u8bc6\u7684\u68c0\u7d22\u3001\u63a8\u7406\u63d0\u4f9b\u4e86\u66f4\u5927\u7684\u524d\u666f<\/strong><\/p>\n\n\n\n<p>\u50cf ChatGPT \u8fd9\u6837\u7684\u8bed\u8a00\u6a21\u578b\uff0c\u6570\u636e\u53ea\u66f4\u65b0\u5230 2021 \u5e74\uff0c\u5982\u4f55\u8ba9\u5927\u6a21\u578b\u56de\u7b54\u548c\u5b66\u4e60\u5230\u4e4b\u540e\u7684\u77e5\u8bc6\u5c31\u662f\u4e00\u4e2a\u5f88\u91cd\u8981\u7684\u95ee\u9898\u3002\u800c\u4e14 ChatGPT \u7684 API \u662f\u6709\u63d0\u793a\u8bcd\u548c\u8fd4\u56de\u5185\u5bb9\u7684\u9650\u5236\u7684\uff0c3.5 \u662f 4k\uff0c4 \u5219\u662f 8k\uff0c\u800c\u6211\u4eec\u5f80\u5f80\u9700\u8981\u4ece\u81ea\u5df1\u7684\u6570\u636e\u3001\u81ea\u5df1\u7684\u6587\u6863\u4e2d\u83b7\u53d6\u7279\u5b9a\u7684\u4fe1\u606f\uff0c\u8fd9\u53ef\u80fd\u662f\u4e00\u672c\u4e66\u3001\u4e00\u4e2a PDF \u6587\u4ef6\u3001\u4e00\u4e2a\u5e26\u6709\u4e13\u6709\u4fe1\u606f\u7684\u6570\u636e\u5e93\u3002\u8fd9\u4e9b\u4fe1\u606f\u7684 token \u6570\u91cf\u4f1a\u8fdc\u9ad8\u4e8e 4k \u7684\u9608\u503c\uff0c\u76f4\u63a5\u4f7f\u7528\u5927\u6a21\u578b\u662f\u65e0\u6cd5\u83b7\u53d6\u5230\u76f8\u5e94\u7684\u77e5\u8bc6\u7684\uff0c\u56e0\u4e3a\u8d85\u8fc7\u9608\u503c\u7684\u4fe1\u606f\u5c31\u88ab\u622a\u65ad\u4e86\u3002<\/p>\n\n\n\n<p>LangChain \u63d0\u4f9b\u4e86\u5bf9\u5411\u91cf\u6570\u636e\u5e93\u7684\u652f\u6301\uff0c\u80fd\u591f\u628a\u8d85\u957f\u7684 txt\u3001pdf \u7b49\u901a\u8fc7\u5927\u6a21\u578b\u8f6c\u6362\u4e3a embedding \u7684\u5f62\u5f0f\uff0c\u5b58\u5230\u5411\u91cf\u6570\u636e\u5e93\u4e2d\uff0c\u7136\u540e\u5229\u7528\u6570\u636e\u5e93\u8fdb\u884c\u68c0\u7d22\u3002\u8fd9\u6837\u5c31\u53ef\u4ee5\u652f\u6301\u66f4\u591a\u957f\u5ea6\u7684\u8f93\u5165\uff0c\u89e3\u653e\u4e86 LLM \u7684\u4f18\u52bf\u3002<\/p>\n\n\n\n<p><strong>\u4e09\u3001LangChain \u7684\u57fa\u672c\u6982\u5ff5 <strong>\ud83d\udd16<\/strong><\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>LangChain \u80fd\u89e3\u51b3\u5927\u6a21\u578b\u7684\u4e24\u4e2a\u75db\u70b9\uff0c\u5305\u62ec\u6a21\u578b\u63a5\u53e3\u590d\u6742\u3001\u8f93\u5165\u957f\u5ea6\u53d7\u9650\u79bb\u4e0d\u5f00\u81ea\u5df1\u7cbe\u5fc3\u8bbe\u8ba1\u7684\u6a21\u5757\u3002\u6839\u636eLangChain \u7684\u6700\u65b0\u6587\u6863\uff0c\u76ee\u524d\u5728 LangChain \u4e2d\u4e00\u5171\u6709\u516d\u5927\u6838\u5fc3\u7ec4\u4ef6\uff0c\u5206\u522b\u662f\u6a21\u578b\u7684\u8f93\u5165\u8f93\u51fa (Model I\/O)\u3001\u6570\u636e\u8fde\u63a5 (Data Connection)\u3001\u5185\u5b58\u8bb0\u5fc6\uff08Memory\uff09\u3001\u94fe\uff08Chains\uff09\u3001\u4ee3\u7406\uff08Agent\uff09\u3001\u56de\u8c03\uff08Callbacks\uff09\u3002\u4e0b\u9762\u6211\u4eec\u5c06\u5206\u522b\u8bb2\u8ff0\u6bcf\u4e00\u4e2a\u6a21\u5757\u7684\u529f\u80fd\u548c\u4f5c\u7528\u3002<\/p>\n\n\n\n<p>\u76ee\u524d\uff0c\u6700\u65b0\u7684\u5b98\u7f51\u4e2d\u5c06\u6570\u636e\u8fde\u63a5\u90e8\u5206\u6539\u4e3a\u4e86\u68c0\u7d22\uff08Retrieval\uff09\uff0c\u4f46\u57fa\u672c\u5185\u5bb9\u5dee\u5f02\u4e0d\u5927\u3002<\/p>\n\n\n\n<p><strong>\u4e00\uff09Model I\/O<\/strong><\/p>\n\n\n\n<p>\u6a21\u578b\u662f\u4efb\u4f55 LLM \u5e94\u7528\u4e2d\u6700\u6838\u5fc3\u7684\u4e00\u70b9\uff0cLangChain \u53ef\u4ee5\u8ba9\u6211\u4eec\u65b9\u4fbf\u7684\u63a5\u5165\u5404\u79cd\u5404\u6837\u7684\u8bed\u8a00\u6a21\u578b\uff0c\u5e76\u4e14\u63d0\u4f9b\u4e86\u8bb8\u591a\u63a5\u53e3\uff0c\u4e3b\u8981\u6709\u4e09\u4e2a\u7ec4\u4ef6\u7ec4\u6210\uff0c\u5305\u62ec\u6a21\u578b\uff08Models\uff09\uff0c\u63d0\u793a\u8bcd\uff08Prompts\uff09\u548c\u89e3\u6790\u5668\uff08Output parsers\uff09\u3002<\/p>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"377\" src=\"https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-1024x377.png\" alt=\"\" class=\"wp-image-19707\" style=\"width:594px;height:auto\" srcset=\"https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-1024x377.png 1024w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-300x110.png 300w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-768x282.png 768w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-830x305.png 830w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-230x85.png 230w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-350x129.png 350w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55-480x177.png 480w, https:\/\/92it.top\/wp-content\/uploads\/2025\/11\/image-55.png 1376w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure><\/div>\n\n\n<p><strong>1.Models<\/strong><\/p>\n\n\n\n<p>LangChain \u4e2d\u63d0\u4f9b\u4e86\u591a\u79cd\u4e0d\u540c\u7684\u8bed\u8a00\u6a21\u578b\uff0c\u6309\u529f\u80fd\u5212\u5206\uff0c\u4e3b\u8981\u6709\u4e24\u79cd\u3002<\/p>\n\n\n\n<ul>\n<li>\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\uff1a\u6211\u4eec\u901a\u5e38\u8bf4\u7684\u8bed\u8a00\u6a21\u578b\uff0c\u7ed9\u5b9a\u8f93\u5165\u7684\u4e00\u4e2a\u6587\u672c\uff0c\u4f1a\u8fd4\u56de\u4e00\u4e2a\u76f8\u5e94\u7684\u6587\u672c\u3002\u5e38\u89c1\u7684\u8bed\u8a00\u6a21\u578b\u6709 GPT3.5\uff0cchatglm\uff0cGPT4All \u7b49\u3002<\/li>\n<\/ul>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.llms import OpenAI\nllm = OpenAI(openai_api_key=\"...\")\n\n<\/pre>\n\n\n\n<p><br><\/p>\n\n\n\n<ul>\n<li>\u804a\u5929\u6a21\u578b\uff08Chat model\uff09\uff1a\u53ef\u4ee5\u770b\u505a\u662f\u5c01\u88c5\u597d\u7684\u62e5\u6709\u5bf9\u8bdd\u80fd\u529b\u7684 LLM\uff0c\u8fd9\u4e9b\u6a21\u578b\u5141\u8bb8\u4f60\u4f7f\u7528\u5bf9\u8bdd\u7684\u5f62\u5f0f\u548c\u5176\u8fdb\u884c\u4ea4\u4e92\uff0c\u80fd\u591f\u652f\u6301\u5c06\u804a\u5929\u4fe1\u606f\u4f5c\u4e3a\u8f93\u5165\uff0c\u5e76\u8fd4\u56de\u804a\u5929\u4fe1\u606f\u3002\u8fd9\u4e9b\u804a\u5929\u4fe1\u606f\u90fd\u662f\u5c01\u88c5\u597d\u7684\u7ed3\u6784\u4f53\uff0c\u800c\u975e\u4e00\u4e2a\u7b80\u5355\u7684\u6587\u672c\u5b57\u7b26\u4e32\u3002\u5e38\u89c1\u7684\u804a\u5929\u6a21\u578b\u6709 GPT4\u3001Llama \u548c Llama2\uff0c\u4ee5\u53ca\u5fae\u8f6f\u4e91 Azure \u76f8\u5173\u7684 GPT \u6a21\u578b\u3002<\/li>\n<\/ul>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.chat_models import ChatOpenAI\nchat = ChatOpenAI(openai_api_key=\"...\")\n\n<\/pre>\n\n\n\n<p><strong>2.Prompts<br><\/strong>\u63d0\u793a\u8bcd\u662f\u6a21\u578b\u7684\u8f93\u5165\uff0c\u901a\u8fc7\u7f16\u5199\u63d0\u793a\u8bcd\u53ef\u4ee5\u548c\u6a21\u578b\u8fdb\u884c\u4ea4\u4e92\u3002LangChain \u4e2d\u63d0\u4f9b\u4e86\u8bb8\u591a\u6a21\u677f\u548c\u51fd\u6570\u7528\u4e8e\u6a21\u5757\u5316\u6784\u5efa\u63d0\u793a\u8bcd\uff0c\u8fd9\u4e9b\u6a21\u677f\u53ef\u4ee5\u63d0\u4f9b\u66f4\u7075\u6d3b\u7684\u65b9\u6cd5\u53bb\u751f\u6210\u63d0\u793a\u8bcd\uff0c\u5177\u6709\u66f4\u597d\u7684\u590d\u7528\u6027\u3002\u6839\u636e\u8c03\u7528\u7684\u6a21\u578b\u65b9\u5f0f\u4e0d\u540c\uff0c\u63d0\u793a\u8bcd\u6a21\u677f\u4e3b\u8981\u5206\u4e3a\u666e\u901a\u6a21\u677f\u4ee5\u53ca\u804a\u5929\u63d0\u793a\u8bcd\u6a21\u677f\u3002<\/p>\n\n\n\n<p><strong>\u63d0\u793a\u6a21\u677f\uff08PromptTemplate\uff09<\/strong><\/p>\n\n\n\n<ul>\n<li>\u63d0\u793a\u6a21\u677f\u662f\u4e00\u79cd\u751f\u6210\u63d0\u793a\u7684\u65b9\u5f0f\uff0c\u5305\u542b\u4e00\u4e2a\u5e26\u6709\u53ef\u66ff\u6362\u5185\u5bb9\u7684\u6a21\u677f\uff0c\u4ece\u7528\u6237\u90a3\u83b7\u53d6\u4e00\u7ec4\u53c2\u6570\u5e76\u751f\u6210\u63d0\u793a<\/li>\n\n\n\n<li>\u63d0\u793a\u6a21\u677f\u7528\u6765\u751f\u6210 LLMs \u7684\u63d0\u793a\uff0c\u6700\u7b80\u5355\u7684\u4f7f\u7528\u573a\u666f\uff0c\u6bd4\u5982\u201c\u6211\u5e0c\u671b\u4f60\u626e\u6f14\u4e00\u4e2a\u4ee3\u7801\u4e13\u5bb6\u7684\u89d2\u8272\uff0c\u544a\u8bc9\u6211\u8fd9\u4e2a\u65b9\u6cd5\u7684\u539f\u7406 {code}\u201d\u3002<\/li>\n\n\n\n<li>\u7c7b\u4f3c\u4e8e python \u4e2d\u7528\u5b57\u5178\u7684\u65b9\u5f0f\u683c\u5f0f\u5316\u5b57\u7b26\u4e32\uff0c\u4f46\u5728 langchain \u4e2d\u90fd\u88ab\u5c01\u88c5\u6210\u4e86\u5bf9\u8c61<\/li>\n<\/ul>\n\n\n\n<p>\u4e00\u4e2a\u7b80\u5355\u7684\u8c03\u7528\u6837\u4f8b\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain import PromptTemplate\n\n\ntemplate = \"\"\"\\\nYou are a naming consultant for new companies.\nWhat is a good name for a company that makes {product}?\n\"\"\"\n\nprompt = PromptTemplate.from_template(template)\nprompt.format(product=\"colorful socks\")\n\n<\/pre>\n\n\n\n<p>\u8f93\u51fa\u7ed3\u679c\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">csharp\n\u590d\u5236\u4ee3\u7801\n# \u5b9e\u9645\u8f93\u51fa\nYou are a naming consultant for new companies.\nWhat is a good name for a company that makes colorful socks?\n\n<\/pre>\n\n\n\n<p><strong>\u804a\u5929\u63d0\u793a\u6a21\u677f\uff08ChatPromptTemplate\uff09<\/strong><\/p>\n\n\n\n<ul>\n<li>\u804a\u5929\u6a21\u578b\u63a5\u6536\u804a\u5929\u6d88\u606f\u4f5c\u4e3a\u8f93\u5165\uff0c\u8fd9\u4e9b\u804a\u5929\u6d88\u606f\u901a\u5e38\u79f0\u4e3a Message\uff0c\u548c\u539f\u59cb\u7684\u63d0\u793a\u6a21\u677f\u4e0d\u4e00\u6837\u7684\u662f\uff0c\u8fd9\u4e9b\u6d88\u606f\u90fd\u4f1a\u548c\u4e00\u4e2a\u89d2\u8272\u8fdb\u884c\u5173\u8054\u3002<\/li>\n\n\n\n<li>\u5728\u4f7f\u7528\u804a\u5929\u6a21\u578b\u65f6\uff0c\u5efa\u8bae\u4f7f\u7528\u804a\u5929\u63d0\u793a\u8bcd\u6a21\u677f\uff0c\u8fd9\u6837\u53ef\u4ee5\u5145\u5206\u53d1\u6325\u804a\u5929\u6a21\u578b\u7684\u6f5c\u529b\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u4e00\u4e2a\u7b80\u5355\u7684\u4f7f\u7528\u793a\u4f8b\u5982\u4e0b\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.prompts import (\n    ChatPromptTemplate,\n    PromptTemplate,\n    SystemMessagePromptTemplate,\n    AIMessagePromptTemplate,\n    HumanMessagePromptTemplate,\n)\nfrom langchain.schema import (\n    AIMessage,\n    HumanMessage,\n    SystemMessage\n)\ntemplate=\"You are a helpful assistant that translates {input_language} to {output_language}.\"\nsystem_message_prompt = SystemMessagePromptTemplate.from_template(template)\nhuman_template=\"{text}\"\nhuman_message_prompt = HumanMessagePromptTemplate.from_template(human_template)\nchat_prompt = ChatPromptTemplate.from_messages([system_message_prompt, human_message_prompt])\n\n# get a chat completion from the formatted messages\nchat_prompt.format_prompt(input_language=\"English\", output_language=\"French\", text=\"I love programming.\").to_messages()\n\n<\/pre>\n\n\n\n<p>\u8f93\u51fa\u7ed3\u679c\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\n[SystemMessage(content='You are a helpful assistant that translates English to French.', additional_kwargs={}),\n     HumanMessage(content='I love programming.', additional_kwargs={})]\n\n<\/pre>\n\n\n\n<p><strong>3.Output parsers<br><\/strong>\u8bed\u8a00\u6a21\u578b\u8f93\u51fa\u7684\u662f\u666e\u901a\u7684\u5b57\u7b26\u4e32\uff0c\u6709\u7684\u65f6\u5019\u6211\u4eec\u53ef\u80fd\u60f3\u5f97\u5230\u7ed3\u6784\u5316\u7684\u8868\u793a\uff0c\u6bd4\u5982 JSON \u6216\u8005 CSV\uff0c\u4e00\u4e2a\u6709\u6548\u7684\u65b9\u6cd5\u5c31\u662f\u4f7f\u7528\u8f93\u51fa\u89e3\u6790\u5668\u3002<\/p>\n\n\n\n<p>\u8f93\u51fa\u89e3\u6790\u5668\u662f\u5e2e\u52a9\u6784\u5efa\u8bed\u8a00\u6a21\u578b\u8f93\u51fa\u7684\u7c7b\uff0c\u4e3b\u8981\u5b9e\u73b0\u4e86\u4e24\u4e2a\u529f\u80fd\uff1a<\/p>\n\n\n\n<ul>\n<li>\u83b7\u53d6\u683c\u5f0f\u6307\u4ee4\uff0c\u662f\u4e00\u4e2a\u6587\u672c\u5b57\u7b26\u4e32\u9700\u8981\u6307\u660e\u8bed\u8a00\u6a21\u578b\u7684\u8f93\u51fa\u5e94\u8be5\u5982\u4f55\u88ab\u683c\u5f0f\u5316<\/li>\n\n\n\n<li>\u89e3\u6790\uff0c\u4e00\u79cd\u63a5\u53d7\u5b57\u7b26\u4e32\u5e76\u5c06\u5176\u89e3\u6790\u6210\u56fa\u5b9a\u7ed3\u6784\u7684\u65b9\u6cd5\uff0c\u53ef\u4ee5\u81ea\u5b9a\u4e49\u89e3\u6790\u5b57\u7b26\u4e32\u7684\u65b9\u5f0f<\/li>\n<\/ul>\n\n\n\n<p>\u4e00\u4e2a\u7b80\u5355\u7684\u4f7f\u7528\u793a\u4f8b\u5982\u4e0b\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.prompts import PromptTemplate, ChatPromptTemplate, HumanMessagePromptTemplate\nfrom langchain.llms import OpenAI\nfrom langchain.chat_models import ChatOpenAI\n\nfrom langchain.output_parsers import PydanticOutputParser\nfrom pydantic import BaseModel, Field, validator\nfrom typing import List\n\nmodel_name = 'text-davinci-003'\ntemperature = 0.0\nmodel = OpenAI(model_name=model_name, temperature=temperature)\n# Define your desired data structure.\nclass Joke(BaseModel):\n    setup: str = Field(description=\"question to set up a joke\")\n    punchline: str = Field(description=\"answer to resolve the joke\")\n    \n    # You can add custom validation logic easily with Pydantic.\n    @validator('setup')\n    def question_ends_with_question_mark(cls, field):\n        if field[-1] != '?':\n            raise ValueError(\"Badly formed question!\")\n        return field\n # Set up a parser + inject instructions into the prompt template.\nparser = PydanticOutputParser(pydantic_object=Joke)\nprompt = PromptTemplate(\n    template=\"Answer the user query.\\n{format_instructions}\\n{query}\\n\",\n    input_variables=[\"query\"],\n    partial_variables={\"format_instructions\": parser.get_format_instructions()}\n)\n# And a query intended to prompt a language model to populate the data structure.\njoke_query = \"Tell me a joke.\"\n_input = prompt.format_prompt(query=joke_query)\noutput = model(_input.to_string())\nparser.parse(output)\n\n<\/pre>\n\n\n\n<p>\u8f93\u51fa\u7ed3\u679c\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nJoke(setup='Why did the chicken cross the road?', punchline='To get to the other side!')\n\n<\/pre>\n\n\n\n<p><strong>\u4e8c\uff09Data Connection<\/strong><\/p>\n\n\n\n<p>\u6709\u7684\u65f6\u5019\uff0c\u6211\u4eec\u5e0c\u671b\u8bed\u8a00\u6a21\u578b\u53ef\u4ee5\u4ece\u81ea\u5df1\u7684\u6570\u636e\u4e2d\u8fdb\u884c\u67e5\u8be2\uff0c\u800c\u4e0d\u662f\u4ec5\u4f9d\u9760\u81ea\u5df1\u672c\u8eab\u8f93\u51fa\u4e00\u4e2a\u7ed3\u679c\u3002\u6570\u636e\u8fde\u63a5\u5668\u7684\u7ec4\u4ef6\u5c31\u5141\u8bb8\u4f60\u4f7f\u7528\u5185\u7f6e\u7684\u65b9\u6cd5\u53bb\u8bfb\u53d6\u3001\u4fee\u6539\uff0c\u5b58\u50a8\u548c\u67e5\u8be2\u81ea\u5df1\u7684\u6570\u636e\uff0c\u4e3b\u8981\u6709\u4e0b\u9762\u51e0\u4e2a\u7ec4\u4ef6\u7ec4\u6210\u3002<\/p>\n\n\n\n<ul>\n<li>\u6587\u6863\u52a0\u8f7d\u5668\uff08Document loaders\uff09\uff1a\u8fde\u63a5\u4e0d\u540c\u7684\u6570\u636e\u6e90\uff0c\u52a0\u8f7d\u6587\u6863\u3002<\/li>\n\n\n\n<li>\u6587\u6863\u8f6c\u6362\u5668\uff08Document transformers\uff09\uff1a\u5b9a\u4e49\u4e86\u5e38\u89c1\u7684\u4e00\u4e9b\u5bf9\u6587\u6863\u52a0\u5de5\u7684\u64cd\u4f5c\uff0c\u6bd4\u5982\u5207\u5206\u6587\u6863\uff0c\u4e22\u5f03\u65e0\u7528\u7684\u6570\u636e<\/li>\n\n\n\n<li>\u6587\u672c\u5411\u91cf\u6a21\u578b\uff08Text embedding models\uff09\uff1a\u5c06\u975e\u7ed3\u6784\u5316\u7684\u6587\u672c\u6570\u636e\u8f6c\u6362\u6210\u4e00\u4e2a\u56fa\u5b9a\u7ef4\u5ea6\u7684\u6d6e\u70b9\u6570\u5411\u91cf<\/li>\n\n\n\n<li>\u5411\u91cf\u6570\u636e\u5e93\uff08Vector stores\uff09\uff1a\u5b58\u50a8\u548c\u68c0\u7d22\u4f60\u7684\u5411\u91cf\u6570\u636e<\/li>\n\n\n\n<li>\u68c0\u7d22\u5668\uff08Retrievers\uff09\uff1a\u7528\u4e8e\u68c0\u7d22\u4f60\u7684\u6570\u636e<\/li>\n<\/ul>\n\n\n\n<p><strong>\u4e09\uff09Chains<\/strong><\/p>\n\n\n\n<p>\u53ea\u4f7f\u7528\u4e00\u4e2a LLM \u53bb\u5f00\u53d1\u5e94\u7528\uff0c\u6bd4\u5982\u804a\u5929\u673a\u5668\u4eba\u662f\u5f88\u7b80\u5355\u7684\uff0c\u4f46\u66f4\u591a\u7684\u65f6\u5019\uff0c\u6211\u4eec\u9700\u8981\u7528\u5230\u8bb8\u591a LLM \u53bb\u5171\u540c\u5b8c\u6210\u4e00\u4e2a\u4efb\u52a1\uff0c\u8fd9\u6837\u539f\u6765\u7684\u6a21\u5f0f\u5c31\u4e0d\u8db3\u4ee5\u652f\u6491\u8fd9\u79cd\u590d\u6742\u7684\u5e94\u7528\u3002<\/p>\n\n\n\n<p>\u4e3a\u6b64 LangChain \u63d0\u51fa\u4e86 Chain \u8fd9\u4e2a\u6982\u5ff5\uff0c\u4e5f\u5c31\u662f\u4e00\u4e2a\u6240\u6709\u7ec4\u4ef6\u7684\u5e8f\u5217\uff0c\u80fd\u591f\u628a\u4e00\u4e2a\u4e2a\u72ec\u7acb\u7684 LLM \u94fe\u63a5\u6210\u4e00\u4e2a\u7ec4\u4ef6\uff0c\u4ece\u800c\u53ef\u4ee5\u5b8c\u6210\u66f4\u590d\u6742\u7684\u4efb\u52a1\u3002\u4e3e\u4e2a\u4f8b\u5b50\uff0c\u6211\u4eec\u53ef\u4ee5\u521b\u5efa\u4e00\u4e2a chain\uff0c\u7528\u4e8e\u63a5\u6536\u7528\u6237\u7684\u8f93\u5165\uff0c\u7136\u540e\u4f7f\u7528\u63d0\u793a\u8bcd\u6a21\u677f\u5c06\u5176\u683c\u5f0f\u5316\uff0c\u6700\u540e\u5c06\u683c\u5f0f\u5316\u7684\u7ed3\u679c\u8f93\u51fa\u5230\u4e00\u4e2a LLM\u3002\u901a\u8fc7\u8fd9\u79cd\u94fe\u5f0f\u7684\u7ec4\u5408\uff0c\u5c31\u53ef\u4ee5\u6784\u6210\u66f4\u591a\u66f4\u590d\u6742\u7684 chain\u3002<\/p>\n\n\n\n<p>\u5728 LangChain \u4e2d\u6709\u8bb8\u591a\u5b9e\u73b0\u597d\u7684 chain\uff0c\u4ee5\u6700\u57fa\u7840\u7684 LLMChain \u4e3a\u4f8b\uff0c\u5b83\u4e3b\u8981\u5b9e\u73b0\u7684\u5c31\u662f\u63a5\u6536\u4e00\u4e2a\u63d0\u793a\u8bcd\u6a21\u677f\uff0c\u7136\u540e\u5bf9\u7528\u6237\u8f93\u5165\u8fdb\u884c\u683c\u5f0f\u5316\uff0c\u7136\u540e\u8f93\u5165\u5230\u4e00\u4e2a LLM\uff0c\u6700\u7ec8\u8fd4\u56de LLM \u7684\u8f93\u51fa\u3002<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.llms import OpenAI\nfrom langchain.prompts import PromptTemplate\n\nllm = OpenAI(temperature=0.9)\nprompt = PromptTemplate(\n    input_variables=[\"product\"],\n    template=\"What is a good name for a company that makes {product}?\",\n)\n\nfrom langchain.chains import LLMChain\nchain = LLMChain(llm=llm, prompt=prompt)\n\n# Run the chain only specifying the input variable.\nprint(chain.run(\"colorful socks\"))\n\n<\/pre>\n\n\n\n<p>LLMChain \u4e0d\u4ec5\u652f\u6301 llm\uff0c\u540c\u6837\u4e5f\u652f\u6301 chat llm\uff0c\u4e0b\u9762\u662f\u4e00\u4e2a\u8c03\u7528\u793a\u4f8b\uff1a<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.chat_models import ChatOpenAI\nfrom langchain.prompts.chat import (\n    ChatPromptTemplate,\n    HumanMessagePromptTemplate,\n)\nhuman_message_prompt = HumanMessagePromptTemplate(\n        prompt=PromptTemplate(\n            template=\"What is a good name for a company that makes {product}?\",\n            input_variables=[\"product\"],\n        )\n    )\nchat_prompt_template = ChatPromptTemplate.from_messages([human_message_prompt])\nchat = ChatOpenAI(temperature=0.9)\nchain = LLMChain(llm=chat, prompt=chat_prompt_template)\nprint(chain.run(\"colorful socks\"))\n\n<\/pre>\n\n\n\n<p><strong>\u56db\uff09Memory<\/strong><\/p>\n\n\n\n<p>\u5927\u591a\u6570\u7684 LLM \u5e94\u7528\u7a0b\u5e8f\u90fd\u4f1a\u6709\u4e00\u4e2a\u4f1a\u8bdd\u63a5\u53e3\uff0c\u5141\u8bb8\u6211\u4eec\u548c LLM \u8fdb\u884c\u591a\u8f6e\u7684\u5bf9\u8bdd\uff0c\u5e76\u6709\u4e00\u5b9a\u7684\u4e0a\u4e0b\u6587\u8bb0\u5fc6\u80fd\u529b\u3002\u4f46\u5b9e\u9645\u4e0a\uff0c\u6a21\u578b\u672c\u8eab\u662f\u4e0d\u4f1a\u8bb0\u5fc6\u4efb\u4f55\u4e0a\u4e0b\u6587\u7684\uff0c\u53ea\u80fd\u4f9d\u9760\u7528\u6237\u672c\u8eab\u7684\u8f93\u5165\u53bb\u4ea7\u751f\u8f93\u51fa\u3002\u800c\u5b9e\u73b0\u8fd9\u4e2a\u8bb0\u5fc6\u529f\u80fd\uff0c\u5c31\u9700\u8981\u989d\u5916\u7684\u6a21\u5757\u53bb\u4fdd\u5b58\u6211\u4eec\u548c\u6a21\u578b\u5bf9\u8bdd\u7684\u4e0a\u4e0b\u6587\u4fe1\u606f\uff0c\u7136\u540e\u5728\u4e0b\u4e00\u6b21\u8bf7\u6c42\u65f6\uff0c\u628a\u6240\u6709\u7684\u5386\u53f2\u4fe1\u606f\u90fd\u8f93\u5165\u7ed9\u6a21\u578b\uff0c\u8ba9\u6a21\u578b\u8f93\u51fa\u6700\u7ec8\u7ed3\u679c\u3002<\/p>\n\n\n\n<p>\u800c\u5728 LangChain \u4e2d\uff0c\u63d0\u4f9b\u8fd9\u4e2a\u529f\u80fd\u7684\u6a21\u5757\u5c31\u79f0\u4e3a Memory\uff0c\u7528\u4e8e\u5b58\u50a8\u7528\u6237\u548c\u6a21\u578b\u4ea4\u4e92\u7684\u5386\u53f2\u4fe1\u606f\u3002\u5728 LangChain \u4e2d\u6839\u636e\u529f\u80fd\u548c\u8fd4\u56de\u503c\u7684\u4e0d\u540c\uff0c\u4f1a\u6709\u591a\u79cd\u4e0d\u540c\u7684 Memory \u7c7b\u578b\uff0c\u4e3b\u8981\u53ef\u4ee5\u5206\u4e3a\u4ee5\u4e0b\u51e0\u4e2a\u7c7b\u522b\uff1a<\/p>\n\n\n\n<ul>\n<li>\u5bf9\u8bdd\u7f13\u51b2\u533a\u5185\u5b58\uff08ConversationBufferMemory\uff09\uff0c\u6700\u57fa\u7840\u7684\u5185\u5b58\u6a21\u5757\uff0c\u7528\u4e8e\u5b58\u50a8\u5386\u53f2\u7684\u4fe1\u606f<\/li>\n\n\n\n<li>\u5bf9\u8bdd\u7f13\u51b2\u5668\u7a97\u53e3\u5185\u5b58\uff08ConversationBufferWindowMemory\uff09\uff0c\u53ea\u4fdd\u5b58\u6700\u540e\u7684 K \u8f6e\u5bf9\u8bdd\u7684\u4fe1\u606f\uff0c\u56e0\u6b64\u8fd9\u79cd\u5185\u5b58\u7a7a\u95f4\u4f7f\u7528\u4f1a\u76f8\u5bf9\u8f83\u5c11<\/li>\n\n\n\n<li>\u5bf9\u8bdd\u6458\u8981\u5185\u5b58\uff08ConversationSummaryMemory\uff09\uff0c\u8fd9\u79cd\u6a21\u5f0f\u4f1a\u5bf9\u5386\u53f2\u7684\u6240\u6709\u4fe1\u606f\u8fdb\u884c\u62bd\u53d6\uff0c\u751f\u6210\u6458\u8981\u4fe1\u606f\uff0c\u7136\u540e\u5c06\u6458\u8981\u4fe1\u606f\u4f5c\u4e3a\u5386\u53f2\u4fe1\u606f\u8fdb\u884c\u4fdd\u5b58\u3002<\/li>\n\n\n\n<li>\u5bf9\u8bdd\u6458\u8981\u7f13\u5b58\u5185\u5b58\uff08ConversationSummaryBufferMemory\uff09\uff0c\u8fd9\u4e2a\u548c\u4e0a\u9762\u7684\u4f5c\u7528\u57fa\u672c\u4e00\u81f4\uff0c\u4f46\u662f\u6709\u6700\u5927 token \u6570\u7684\u9650\u5236\uff0c\u8fbe\u5230\u8fd9\u4e2a\u6700\u5927 token \u6570\u7684\u65f6\u5019\u5c31\u4f1a\u8fdb\u884c\u5408\u5e76\u5386\u53f2\u4fe1\u606f\u751f\u6210\u6458\u8981<\/li>\n<\/ul>\n\n\n\n<p>\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u5bf9\u8bdd\u6458\u8981\u5185\u5b58\u7684\u8bbe\u8ba1\u51fa\u53d1\u70b9\u5c31\u662f\u8bed\u8a00\u6a21\u578b\u80fd\u652f\u6301\u7684\u4e0a\u4e0b\u6587\u957f\u5ea6\u662f\u6709\u9650\u7684\uff08\u4e00\u822c\u662f 2048\uff09\uff0c\u8d85\u8fc7\u4e86\u8fd9\u4e2a\u957f\u5ea6\u7684\u6570\u636e\u5929\u7136\u7684\u5c31\u88ab\u622a\u65ad\u4e86\u3002\u8fd9\u4e2a\u7c7b\u4f1a\u6839\u636e\u5bf9\u8bdd\u7684\u8f6e\u6b21\u8fdb\u884c\u5408\u5e76\uff0c\u9ed8\u8ba4\u503c\u662f 2\uff0c\u4e5f\u5c31\u662f\u6bcf 2 \u8f6e\u5c31\u5f00\u542f\u4e00\u6b21\u8c03\u7528 LLM \u53bb\u5408\u5e76\u5386\u53f2\u4fe1\u606f\u3002<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.memory import ConversationBufferMemory\nmemory = ConversationBufferMemory(memory_key=\"chat_history\")\nmemory.chat_memory.add_user_message(\"hi!\")\nmemory.chat_memory.add_ai_message(\"whats up?\")\n\n<\/pre>\n\n\n\n<p>\u53c2\u8003\u5b98\u65b9\u7684\u6559\u7a0b\uff0cMemory \u540c\u65f6\u652f\u6301 LLM \u548c Chat model\u3002<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">python\n\u590d\u5236\u4ee3\u7801\nfrom langchain.llms import OpenAI\nfrom langchain.prompts import PromptTemplate\nfrom langchain.chains import LLMChain\nfrom langchain.memory import ConversationBufferMemory\n\n# llm \nllm = OpenAI(temperature=0)\n# Notice that \"chat_history\" is present in the prompt template\ntemplate = \"\"\"You are a nice chatbot having a conversation with a human.\n\nPrevious conversation:\n{chat_history}\n\nNew human question: {question}\nResponse:\"\"\"\nprompt = PromptTemplate.from_template(template)\n# Notice that we need to align the `memory_key`\nmemory = ConversationBufferMemory(memory_key=\"chat_history\")\nconversation = LLMChain(\n    llm=llm,\n    prompt=prompt,\n    verbose=True,\n    memory=memory\n)\nconversation({\"question\": \"hi\"})\n\n<\/pre>\n\n\n\n<p>\u4e0b\u9762\u662f\u4f7f\u7528 Chat model \u8c03\u7528 Memory \u7684\u793a\u4f8b\u3002<\/p>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">ini\n\u590d\u5236\u4ee3\u7801\nfrom langchain.chat_models import ChatOpenAI\nfrom langchain.prompts import (\n    ChatPromptTemplate,\n    MessagesPlaceholder,\n    SystemMessagePromptTemplate,\n    HumanMessagePromptTemplate,\n)\nfrom langchain.chains import LLMChain\nfrom langchain.memory import ConversationBufferMemory\n\n\nllm = ChatOpenAI()\nprompt = ChatPromptTemplate(\n    messages=[\n        SystemMessagePromptTemplate.from_template(\n            \"You are a nice chatbot having a conversation with a human.\"\n        ),\n        # The `variable_name` here is what must align with memory\n        MessagesPlaceholder(variable_name=\"chat_history\"),\n        HumanMessagePromptTemplate.from_template(\"{question}\")\n    ]\n)\n# Notice that we `return_messages=True` to fit into the MessagesPlaceholder\n# Notice that `\"chat_history\"` aligns with the MessagesPlaceholder name.\nmemory = ConversationBufferMemory(memory_key=\"chat_history\", return_messages=True)\nconversation = LLMChain(\n    llm=llm,\n    prompt=prompt,\n    verbose=True,\n    memory=memory\n)\nconversation({\"question\": \"hi\"})\n\n<\/pre>\n\n\n\n<p><strong>\u4e94\uff09Agents<br><\/strong>\u4ee3\u7406\u7684\u6838\u5fc3\u601d\u60f3\u5c31\u662f\u4f7f\u7528 LLM \u53bb\u9009\u62e9\u5bf9\u7528\u6237\u7684\u8f93\u5165\uff0c\u5e94\u8be5\u4f7f\u7528\u54ea\u4e2a\u7279\u5b9a\u7684\u5de5\u5177\u53bb\u8fdb\u884c\u64cd\u4f5c\u3002\u8fd9\u91cc\u7684\u5de5\u5177\u53ef\u4ee5\u662f\u53e6\u5916\u7684\u4e00\u4e2a LLM\uff0c\u4e5f\u53ef\u4ee5\u662f\u4e00\u4e2a\u51fd\u6570\u6216\u8005\u4e00\u4e2a chain\u3002\u5728\u4ee3\u7406\u6a21\u5757\u4e2d\uff0c\u6709\u4e09\u4e2a\u6838\u5fc3\u7684\u6982\u5ff5\u3002<\/p>\n\n\n\n<p>1\u3001\u4ee3\u7406\uff08Agent\uff09\uff0c\u4f9d\u6258\u4e8e\u5f3a\u529b\u7684\u8bed\u8a00\u6a21\u578b\u548c\u63d0\u793a\u8bcd\uff0c\u4ee3\u7406\u662f\u7528\u6765\u51b3\u5b9a\u4e0b\u4e00\u6b65\u8981\u505a\u4ec0\u4e48\uff0c\u5176\u6838\u5fc3\u4e5f\u662f\u6784\u5efa\u4e00\u4e2a\u4f18\u79c0\u7684\u63d0\u793a\u8bcd\u3002\u8fd9\u4e2a\u63d0\u793a\u8bcd\u5927\u81f4\u6709\u4e0b\u9762\u51e0\u4e2a\u4f5c\u7528\uff1a<\/p>\n\n\n\n<ul>\n<li>\u89d2\u8272\u5b9a\u4e49\uff0c\u7ed9\u4ee3\u7406\u8bbe\u5b9a\u4e00\u4e2a\u7b26\u5408\u81ea\u5df1\u7684\u8eab\u4efd<\/li>\n\n\n\n<li>\u4e0a\u4e0b\u6587\u4fe1\u606f\uff0c\u63d0\u4f9b\u7ed9\u4ed6\u66f4\u591a\u7684\u4fe1\u606f\u6765\u8981\u6c42\u4ed6\u53ef\u4ee5\u6267\u884c\u4ec0\u4e48\u4efb\u52a1<\/li>\n\n\n\n<li>\u4e30\u5bcc\u7684\u63d0\u793a\u7b56\u7565\uff0c\u589e\u52a0\u4ee3\u7406\u7684\u63a8\u7406\u80fd\u529b<\/li>\n<\/ul>\n\n\n\n<p>2\u3001\u5de5\u5177\uff08Tools\uff09\uff0c\u4ee3\u7406\u4f1a\u9009\u62e9\u4e0d\u540c\u7684\u5de5\u5177\u53bb\u6267\u884c\u4e0d\u540c\u7684\u4efb\u52a1\u3002\u5de5\u5177\u4e3b\u8981\u7ed9\u4ee3\u7406\u63d0\u4f9b\u8c03\u7528\u81ea\u5df1\u7684\u65b9\u6cd5\uff0c\u5e76\u4e14\u4f1a\u63cf\u8ff0\u81ea\u5df1\u5982\u4f55\u88ab\u4f7f\u7528\u3002\u5de5\u5177\u7684\u8fd9\u4e24\u70b9\u90fd\u5341\u5206\u91cd\u8981\uff0c\u5982\u679c\u4f60\u6ca1\u6709\u63d0\u4f9b\u53ef\u4ee5\u8c03\u7528\u5de5\u5177\u7684\u65b9\u6cd5\uff0c\u90a3\u4e48\u4ee3\u7406\u5c31\u6c38\u8fdc\u5b8c\u4e0d\u6210\u81ea\u5df1\u7684\u4efb\u52a1\uff1b\u540c\u65f6\u5982\u679c\u6ca1\u6709\u6b63\u786e\u7684\u63cf\u8ff0\u5de5\u5177\uff0c\u4ee3\u7406\u5c31\u4e0d\u77e5\u9053\u5982\u4f55\u53bb\u4f7f\u7528\u5de5\u5177\u3002<\/p>\n\n\n\n<p>3\u3001\u5de5\u5177\u5305\uff08Toolkits\uff09\uff0cLangChain \u63d0\u4f9b\u4e86\u5de5\u5177\u5305\u7684\u4f7f\u7528\uff0c\u5728\u4e00\u4e2a\u5de5\u5177\u5305\u91cc\u901a\u5e38\u5305\u542b 3-5 \u4e2a\u5de5\u5177\u3002<\/p>\n\n\n\n<p>Agent \u6280\u672f\u662f\u76ee\u524d\u5927\u8bed\u8a00\u6a21\u578b\u7814\u7a76\u7684\u4e00\u4e2a\u524d\u6cbf\u548c\u70ed\u70b9\u65b9\u5411\uff0c\u4f46\u662f\u76ee\u524d\u53d7\u9650\u4e8e\u5927\u6a21\u578b\u7684\u5b9e\u9645\u6548\u679c\uff0c\u4ec5 GPT 4.0 \u53ef\u4ee5\u6709\u6548\u7684\u5f00\u5c55 Agent \u76f8\u5173\u7684\u7814\u7a76\u3002\u6211\u4eec\u76f8\u4fe1\u5728\u672a\u6765\uff0c\u968f\u7740\u5927\u6a21\u578b\u6027\u80fd\u7684\u4f18\u5316\u548c\u8fed\u4ee3\uff0cAgent \u6280\u672f\u5e94\u8be5\u80fd\u6709\u66f4\u597d\u7684\u53d1\u5c55\u548c\u524d\u666f\u3002<\/p>\n\n\n\n<p><strong>\u516d\uff09Callbacks<br><\/strong>\u56de\u8c03\uff0c\u5b57\u9762\u89e3\u91ca\u662f\u8ba9\u7cfb\u7edf\u56de\u8fc7\u6765\u8c03\u7528\u6211\u4eec\u6307\u5b9a\u597d\u7684\u51fd\u6570\u3002\u5728 LangChain \u4e2d\u5c31\u63d0\u4f9b\u4e86\u4e00\u4e2a\u8fd9\u6837\u7684\u56de\u8c03\u7cfb\u7edf\uff0c\u5141\u8bb8\u4f60\u8fdb\u884c\u65e5\u5fd7\u7684\u6253\u5370\u3001\u76d1\u63a7\uff0c\u4ee5\u53ca\u6d41\u5f0f\u4f20\u8f93\u7b49\u5176\u4ed6\u4efb\u52a1\u3002\u901a\u8fc7\u76f4\u63a5\u5728 API \u4e2d\u63d0\u4f9b\u7684\u56de\u8c03\u53c2\u6570\uff0c\u5c31\u53ef\u4ee5\u7b80\u5355\u7684\u5b9e\u73b0\u56de\u8c03\u7684\u529f\u80fd\u3002LangChain \u5185\u7f6e\u4e86\u8bb8\u591a\u53ef\u4ee5\u5b9e\u73b0\u56de\u8c03\u529f\u80fd\u7684\u5bf9\u8c61\uff0c\u6211\u4eec\u901a\u5e38\u79f0\u4e3a handlers\uff0c\u7528\u4e8e\u5b9a\u4e49\u5728\u4e0d\u540c\u4e8b\u4ef6\u89e6\u53d1\u7684\u65f6\u5019\u53ef\u4ee5\u5b9e\u73b0\u7684\u529f\u80fd\u3002<\/p>\n\n\n\n<p>\u4e0d\u7ba1\u4f7f\u7528 Chains\u3001Models\u3001Tools\u3001Agents\uff0c\u53bb\u8c03\u7528 handlers\uff0c\u5747\u901a\u8fc7\u662f\u4f7f\u7528 callbacks \u53c2\u6570\uff0c\u8fd9\u4e2a\u53c2\u6570\u53ef\u4ee5\u5728\u4e24\u4e2a\u4e0d\u540c\u7684\u5730\u65b9\u8fdb\u884c\u4f7f\u7528\uff1a<\/p>\n\n\n\n<ul>\n<li>\u6784\u9020\u51fd\u6570\u4e2d\uff0c\u4f46\u5b83\u7684\u4f5c\u7528\u57df\u53ea\u80fd\u662f\u8be5\u5bf9\u8c61\u3002\u6bd4\u5982\u4e0b\u9762\u8fd9\u4e2a LLMChain \u7684\u6784\u9020\u51fd\u6570\u53ef\u4ee5\u8fdb\u884c\u56de\u8c03\uff0c\u4f46\u8fd9\u4e2a\u56de\u8c03\u51fd\u6570\u5bf9\u4e8e\u94fe\u63a5\u5230\u5b83\u7684 LLM \u6a21\u578b\u662f\u4e0d\u751f\u6548\u7684\u3002<\/li>\n<\/ul>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">css\n\u590d\u5236\u4ee3\u7801\nLLMChain(callbacks=[handler], tags=['a-tag'])\n\n<\/pre>\n\n\n\n<ul>\n<li>\u5728 run()\/apply() \u65b9\u6cd5\u4e2d\u8c03\u7528\uff0c\u53ea\u6709\u5f53\u524d\u8fd9\u4e00\u6b21\u8bf7\u6c42\u624d\u4f1a\u76f8\u5e94\u8fd9\u4e2a\u56de\u8c03\u51fd\u6570\uff0c\u4f46\u662f\u5f53\u524d\u8bf7\u6c42\u5305\u542b\u7684\u5b50\u8bf7\u6c42\u90fd\u4f1a\u8c03\u7528\u8fd9\u4e2a\u56de\u8c03\u3002\u6bd4\u5982\uff0c\u4f7f\u7528\u4e86\u4e00\u4e2a chain \u53bb\u89e6\u53d1\u8fd9\u4e2a\u8bf7\u6c42\uff0c\u8fde\u63a5\u5230\u5b83\u7684 LLM \u6a21\u578b\u4e5f\u4f1a\u8c03\u7528\u8fd9\u4e2a\u56de\u8c03\u3002<\/li>\n<\/ul>\n\n\n\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">css\n\u590d\u5236\u4ee3\u7801\nchain.run(input, callbacks=[handler])\n\n<\/pre>\n\n\n\n<p><strong>\u56db\u3001LangChain \u7684\u4f18\u52bf\ud83d\udd16<\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u548c LangChain \u7c7b\u4f3c\u7684 LLM \u5e94\u7528\u5f00\u53d1\u6846\u67b6\uff1a<\/p>\n\n\n\n<ul>\n<li>OpenAI \u7684 GPT-3.5\/4 API<\/li>\n\n\n\n<li>Hugging Face \u7684 Transformers\uff08\u591a\u6a21\u6001\u673a\u5668\u5b66\u4e60\u6a21\u578b\uff0c\u652f\u6301\u4e0a\u5343\u9884\u8bad\u7ec3\u6a21\u578b\uff09<\/li>\n\n\n\n<li>Google \u7684 T5\uff08NLP \u6846\u67b6\uff09\u7b49<\/li>\n<\/ul>\n\n\n\n<p>LangChain \u7684\u4f18\u52bf\uff1a<\/p>\n\n\n\n<p>\u80fd\u529b\u66f4\u5f3a\uff0c\u66f4\u65b0 by days<\/p>\n\n\n\n<ul>\n<li>\u4ee3\u7801\u8bbe\u8ba1\u4f18\u96c5\uff0c\u6a21\u5757\u5316\u7a0b\u5ea6\u9ad8\uff0cChain\u3001Agent\u3001Memory \u6a21\u5757\u7684\u62bd\u8c61\u7a0b\u5ea6\u9ad8\uff0c\u4fbf\u4e8e\u7ed3\u5408\u5e94\u7528<br>\u96c6\u6210\u5de5\u5177\u5b8c\u5584\uff0c\u4ece\u6570\u636e\u9884\u5904\u7406\u3001LLM \u6a21\u578b\u3001\u5411\u91cf\u5316\u3001\u56fe\u6570\u636e\u5e93\u7b49<br>\u652f\u6301\u5e38\u7528 LLM \u548c\u5927\u91cf\u5546\u4e1a\u5316 NLP \u6a21\u578b<\/li>\n\n\n\n<li>\u5546\u4e1a\u5316\uff1aAzure OpenAI\u3001OpenAI<br>\u5f00\u6e90\uff1aHugging Face\u3001GPT4All<br>\u6709\u5927\u91cf\u7684 LLM \u7528\u4f8b\u4f9b\u53c2\u8003<br><\/li>\n<\/ul>\n\n\n\n<p><strong>\u4e94\u3001\u57fa\u4e8e LangChain \u7684\u5e94\u7528\ud83d\udd16<\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u4ece\u4e0a\u6587\u4e2d\uff0c\u6211\u4eec\u4e86\u89e3\u4e86 LangChain \u7684\u57fa\u672c\u6982\u5ff5\uff0c\u4ee5\u53ca\u4e3b\u8981\u7684\u7ec4\u4ef6\uff0c\u5229\u7528\u8fd9\u4e9b\u80fd\u5e2e\u52a9\u6211\u4eec\u5feb\u901f\u4e0a\u624b\u6784\u5efa app\u3002LangChain \u80fd\u591f\u5728\u5f88\u591a\u4f7f\u7528\u573a\u666f\u4e2d\u8fdb\u884c\u5e94\u7528\uff0c\u5305\u62ec\u4f46\u4e0d\u9650\u4e8e\uff1a<\/p>\n\n\n\n<p>\u4e2a\u4eba\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff1b\u80fd\u591f\u8bb0\u4f4f\u548c\u4f60\u7684\u6bcf\u4e00\u6b21\u4e92\u52a8\uff0c\u5e76\u8fdb\u884c\u4e2a\u6027\u5316\u7684\u4ea4\u4e92<br>\u57fa\u4e8e\u6587\u6863\u7684\u95ee\u7b54\u7cfb\u7edf\uff1b\u5728\u7279\u5b9a\u6587\u6863\u4e0a\u56de\u7b54\u95ee\u9898\uff0c\u53ef\u4ee5\u51cf\u5c11\u5927\u6a21\u578b\u7684\u5e7b\u89c9\u95ee\u9898<br>\u8868\u683c\u6570\u636e\u67e5\u8be2\uff1b\u63d0\u4f9b\u4e86\u5bf9\u7ed3\u6784\u5316\u6570\u636e\u7684\u67e5\u8be2\u529f\u80fd\uff0c\u5982 CSV\uff0cPDF\uff0cSQL\uff0cDataFrame \u7b49<br>API \u4ea4\u4e92\uff1b\u53ef\u4ee5\u5bf9\u63a5\u4e0d\u540c\u8bed\u8a00\u6a21\u578b\u7684API\uff0c\u5e76\u4ea7\u751f\u4ea4\u4e92\u548c\u8c03\u7528<br>\u4fe1\u606f\u63d0\u53d6\uff1b\u4ece\u6587\u672c\u4e2d\u63d0\u53d6\u7ed3\u6784\u5316\u7684\u4fe1\u606f\uff0c\u5e76\u8f93\u51fa<br>\u6587\u6863\u603b\u7ed3\uff1b\u5229\u7528 LLM \u548c embedding \u5bf9\u957f\u6587\u6863\u8fdb\u884c\u538b\u7f29\u548c\u603b\u7ed3<\/p>\n\n\n\n<p>\u800c\u4e14\u5728 github \u4e0a\u4e5f\u6709\u5f88\u591a\u4eba\u5f00\u6e90\u4e86\u57fa\u4e8e LangChain \u5f00\u53d1\u7684\u5f00\u6e90\u5e94\u7528:<\/p>\n\n\n\n<ul>\n<li><a href=\"https:\/\/link.juejin.cn\/?target=https%3A%2F%2Fgithub.com%2Fmayooear%2Fgpt4-pdf-chatbot-langchain\">gpt4-pdf-chatbot<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/link.juejin.cn\/?target=https%3A%2F%2Fgithub.com%2Fgabacode%2FchatPDF\">chatPDF<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/link.juejin.cn\/?target=https%3A%2F%2Fgithub.com%2Fchatchat-space%2FLangchain-Chatchat\">Langchain-Chatchat<\/a><\/li>\n<\/ul>\n\n\n\n<p><strong>\u516d\u3001LangChain \u7684\u7f3a\u70b9\ud83d\udd16<\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u4ece\u5b9e\u9645\u4f7f\u7528\u4f53\u9a8c\u6765\u8bb2\uff0c\u8fd9\u5e76\u4e0d\u662f\u4e00\u4e2a\u5b8c\u7f8e\u7684\u6846\u67b6\uff0c\u4e5f\u5b58\u5728\u4e0d\u5c11\u95ee\u9898\u3002<\/p>\n\n\n\n<p>\u6bd4\u5982\uff0cLangChain \u7684\u63d0\u793a\u8bcd\u6a21\u677f\u5176\u5b9e\u5c31\u662f\u5c01\u88c5\u4e86\u5b57\u7b26\u4e32\u7684\u7c7b\uff0c\u4f46\u662f LangChain \u4e2d\u6709\u5f88\u591a\u7c7b\u578b\u7684\u63d0\u793a\u8bcd\u6a21\u677f\uff0c\u6ca1\u6709\u770b\u51fa\u660e\u663e\u533a\u522b\uff0c\u800c\u4e14\u4e5f\u6ca1\u6709\u5b89\u5168\u6027\uff0c\u5197\u4f59\u6bd4\u8f83\u591a\u3002\u800c\u4e14\u6709\u4e00\u4e9b\u63d0\u5347\u8bcd\u662f\u9ed8\u8ba4\u5199\u597d\u7684\uff0c\u8981\u4fee\u6539\u7684\u8bdd\uff0c\u9700\u8981\u770b\u6e90\u7801\u624d\u77e5\u9053\u5e94\u8be5\u4fee\u6539\u4ec0\u4e48\u5730\u65b9\u3002<\/p>\n\n\n\n<p>LangChain \u5185\u90e8\u5c01\u88c5\u4e86\u5f88\u591a\u8c03\u7528\u8fc7\u7a0b\uff0cdebug \u7684\u8fc7\u7a0b\u6bd4\u8f83\u56f0\u96be\u3002\u4e00\u65e6\u51fa\u73b0\u4e86\u95ee\u9898\uff0c\u6392\u67e5\u6240\u82b1\u8d39\u7684\u65f6\u95f4\u53ef\u80fd\u4f1a\u6bd4\u8f83\u957f\u3002<\/p>\n\n\n\n<p>\u4e4b\u524d\u6709\u7206\u51fa\u8fc7 LangChain \u7684\u4ee3\u7801\u5728\u8c03\u7528 python \u53bb\u6267\u884c agent \u7684\u65f6\u5019\u4f1a\u5b58\u5728\u5b89\u5168\u6f0f\u6d1e\uff0c\u6709\u53ef\u80fd\u901a\u8fc7\u6ce8\u5165\u653b\u51fb\u7684\u65b9\u5f0f\u4ea7\u751f\u5371\u9669\u3002\u4f46\u662f\u8fd9\u4e9b\u7c7b\u4f3c\u7684\u6f0f\u6d1e\uff0c\u9700\u8981\u5b98\u65b9\u53bb\u4fee\u590d\u624d\u53ef\u4ee5\uff0c\u4f1a\u7ed9\u6211\u4eec\u5f00\u53d1\u5e26\u6765\u4e0d\u5fc5\u8981\u7684\u9ebb\u70e6\u3002<\/p>\n\n\n\n<p>LangChain \u7684\u6587\u6863\u8fc7\u4e8e\u7b80\u5355\u4e86\uff0c\u5982\u679c\u8981\u5b9e\u73b0\u4e00\u4e9b\u5b98\u65b9\u6ca1\u6709\u63d0\u4f9b\u7684\u65b9\u6cd5\u5c31\u9700\u8981\u52a8\u4e00\u4e9b\u8111\u7b4b\u3002<\/p>\n\n\n\n<p><strong>\u603b\u7ed3\ud83d\udd16<\/strong><\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p><strong><br><\/strong>LangChain \u4f5c\u4e3a\u4e00\u4e2a\u65b0\u5174\u7684\u5f00\u6e90 LLM \u5f00\u53d1\u6846\u67b6\uff0c\u5b83\u7684\u8bbe\u8ba1\u7406\u5ff5\u548c\u4e00\u4e9b\u5b9e\u73b0\u65b9\u6cd5\u90fd\u6709\u503c\u5f97\u6211\u4eec\u501f\u9274\u7684\u5730\u65b9\u3002\u5f53\u7136\u4e00\u4e2a\u5f00\u6e90\u9879\u76ee\u5fc5\u7136\u6709\u81ea\u5df1\u7684\u4f18\u70b9\u7684\u540c\u65f6\u4e5f\u4f1a\u5b58\u5728\u4e0d\u5c11\u7684\u7f3a\u70b9\u3002\u6211\u4eec\u5728\u5b9e\u9645\u4f7f\u7528\u7684\u65f6\u5019\uff0c\u5e94\u8be5\u6309\u7167\u81ea\u5df1\u7684\u9700\u6c42\u548c\u5b9e\u9645\u60c5\u51b5\u53bb\u9009\u62e9\uff0c\u5207\u8bb0\uff0c\u4e0d\u8981\u7b80\u5355\u7684\u56e0\u4e3a\u7f51\u4e0a\u7684\u8bc4\u8bba\u800c\u505a\u51fa\u9009\u62e9\u3002\u6bd5\u7adf\uff0c\u597d\u4e0d\u597d\u7528\uff0c\u53ea\u6709\u4f60\u81ea\u5df1\u624d\u6e05\u695a\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8f6c\u8f7d\uff1ahttps:\/\/blog.csdn.net\/2401_85375151\/article\/details\/ [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[48],"tags":[],"_links":{"self":[{"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/posts\/19705"}],"collection":[{"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/92it.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=19705"}],"version-history":[{"count":7,"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/posts\/19705\/revisions"}],"predecessor-version":[{"id":19713,"href":"https:\/\/92it.top\/index.php?rest_route=\/wp\/v2\/posts\/19705\/revisions\/19713"}],"wp:attachment":[{"href":"https:\/\/92it.top\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=19705"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/92it.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=19705"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/92it.top\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=19705"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}