{"id":18186,"date":"2024-08-25T00:45:10","date_gmt":"2024-08-24T16:45:10","guid":{"rendered":"https:\/\/17aitech.com\/?p=18186"},"modified":"2024-10-08T15:08:49","modified_gmt":"2024-10-08T07:08:49","slug":"%e3%80%90%e8%af%be%e7%a8%8b%e6%80%bb%e7%bb%93%e3%80%91day28%ef%bc%9a%e5%a4%a7%e6%a8%a1%e5%9e%8b%e4%b9%8b%e6%b7%b1%e5%85%a5%e6%8e%a2%e7%b4%a2rag%e6%b5%81%e7%a8%8b","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=18186","title":{"rendered":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b"},"content":{"rendered":"<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_85 ez-toc-wrap-left-text counter-hierarchy ez-toc-counter ez-toc-light-blue ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">\u6587\u7ae0\u76ee\u5f55<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Toggle Table of Content\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%89%8D%E8%A8%80\" >\u524d\u8a00<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/17aitech.com\/?p=18186\/#RAG%E6%B5%81%E7%A8%8B%E5%9B%9E%E9%A1%BE\" >RAG\u6d41\u7a0b\u56de\u987e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%96%87%E6%A1%A3%E8%AF%BB%E5%8F%96LOAD\" >\u6587\u6863\u8bfb\u53d6(LOAD)<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/17aitech.com\/?p=18186\/#CSV%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >CSV\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/17aitech.com\/?p=18186\/#CSVLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >CSVLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/17aitech.com\/?p=18186\/#TXT%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >TXT\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/17aitech.com\/?p=18186\/#TXTLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >TXTLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-2\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/17aitech.com\/?p=18186\/#PDF%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >PDF\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/17aitech.com\/?p=18186\/#PyMuPDFLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >PyMuPDFLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-12\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-3\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-13\" href=\"https:\/\/17aitech.com\/?p=18186\/#EXCEL%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >EXCEL\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-14\" href=\"https:\/\/17aitech.com\/?p=18186\/#UnstructuredExcelLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >UnstructuredExcelLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-15\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-4\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-16\" href=\"https:\/\/17aitech.com\/?p=18186\/#PPT%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >PPT\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-17\" href=\"https:\/\/17aitech.com\/?p=18186\/#UnstructuredPowerPointLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >UnstructuredPowerPointLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-18\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-5\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-19\" href=\"https:\/\/17aitech.com\/?p=18186\/#WORD%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >WORD\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-20\" href=\"https:\/\/17aitech.com\/?p=18186\/#UnstructuredWordDocumentLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >UnstructuredWordDocumentLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-21\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-6\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-22\" href=\"https:\/\/17aitech.com\/?p=18186\/#MARKDOWN%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >MARKDOWN\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-23\" href=\"https:\/\/17aitech.com\/?p=18186\/#UnstructuredMarkdownLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >UnstructuredMarkdownLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-24\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-7\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-25\" href=\"https:\/\/17aitech.com\/?p=18186\/#HTML%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >HTML\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-26\" href=\"https:\/\/17aitech.com\/?p=18186\/#UnstructuredHTMLLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >UnstructuredHTMLLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-27\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-8\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-28\" href=\"https:\/\/17aitech.com\/?p=18186\/#SQL%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\" >SQL\u6587\u4ef6\u52a0\u8f7d\u5668<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-29\" href=\"https:\/\/17aitech.com\/?p=18186\/#SQLDatabaseLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >SQLDatabaseLoader\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-30\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-9\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-31\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%96%87%E6%A1%A3%E5%88%87%E5%88%86SPLIT\" >\u6587\u6863\u5207\u5206(SPLIT)<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-32\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%8C%89%E7%85%A7%E5%AD%97%E7%AC%A6%E5%88%87%E5%88%86\" >\u6309\u7167\u5b57\u7b26\u5207\u5206<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-33\" href=\"https:\/\/17aitech.com\/?p=18186\/#CharacterTextSplitter%E7%B1%BB%E8%AF%B4%E6%98%8E\" >CharacterTextSplitter\u7c7b\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-34\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-10\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-35\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E9%80%92%E5%BD%92%E6%96%B9%E5%BC%8F%E5%88%87%E5%88%86\" >\u9012\u5f52\u65b9\u5f0f\u5207\u5206<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-36\" href=\"https:\/\/17aitech.com\/?p=18186\/#RecursiveCharacterTextSplitter%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >RecursiveCharacterTextSplitter\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-37\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-11\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-38\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%8C%89%E7%85%A7Token%E5%88%87%E5%88%86\" >\u6309\u7167Token\u5207\u5206<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-39\" href=\"https:\/\/17aitech.com\/?p=18186\/#TokenTextSplitter%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\" >TokenTextSplitter\u51fd\u6570\u8bf4\u660e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-40\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-12\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-41\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%8C%89%E7%85%A7markdown%E6%A0%87%E9%A2%98%E5%88%87%E5%88%86\" >\u6309\u7167markdown\u6807\u9898\u5207\u5206<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-42\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E7%A4%BA%E4%BE%8B-13\" >\u793a\u4f8b<\/a><\/li><\/ul><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-43\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%90%91%E9%87%8F%E5%8C%96EMBED\" >\u5411\u91cf\u5316(EMBED)<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-44\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%9C%AC%E5%9C%B0%E5%8C%96\" >\u672c\u5730\u5316<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-45\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%9C%A8%E7%BA%BF%E5%8C%96\" >\u5728\u7ebf\u5316<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-46\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E4%BD%BF%E7%94%A8HuggingFaceEmbeddings\" >\u4f7f\u7528HuggingFaceEmbeddings<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-47\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E4%BD%BF%E7%94%A8QianfanEmbeddings\" >\u4f7f\u7528QianfanEmbeddings<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-48\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%8E%9F%E7%90%86%E6%8E%A2%E5%AF%BB\" >\u539f\u7406\u63a2\u5bfb<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-49\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%AD%98%E5%82%A8STORE\" >\u5b58\u50a8(STORE)<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-50\" href=\"https:\/\/17aitech.com\/?p=18186\/#Chroma%E7%AE%80%E4%BB%8B\" >Chroma\u7b80\u4ecb<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-51\" href=\"https:\/\/17aitech.com\/?p=18186\/#Chroma%E4%BD%BF%E7%94%A8%E7%9A%84%E5%87%A0%E7%A7%8D%E6%96%B9%E5%BC%8F\" >Chroma\u4f7f\u7528\u7684\u51e0\u79cd\u65b9\u5f0f<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-52\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E9%80%9A%E8%BF%87Chromadb%E7%9A%84%E6%BA%90%E7%94%9F%E6%8E%A5%E5%8F%A3%E4%BD%BF%E7%94%A8Chroma\" >\u901a\u8fc7Chromadb\u7684\u6e90\u751f\u63a5\u53e3\u4f7f\u7528Chroma<\/a><ul class='ez-toc-list-level-5' ><li class='ez-toc-heading-level-5'><a class=\"ez-toc-link ez-toc-heading-53\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E6%8C%81%E4%B9%85%E5%8C%96%E6%A8%A1%E5%BC%8F\" >\u6301\u4e45\u5316\u6a21\u5f0f<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-5'><a class=\"ez-toc-link ez-toc-heading-54\" href=\"https:\/\/17aitech.com\/?p=18186\/#ClientServer%E6%A8%A1%E5%BC%8F\" >Client\/Server\u6a21\u5f0f<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-55\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E9%80%9A%E8%BF%87Langchain%E7%9A%84%E7%BB%84%E5%90%88%E5%8C%85%E4%BD%BF%E7%94%A8Chroma\" >\u901a\u8fc7Langchain\u7684\u7ec4\u5408\u5305\u4f7f\u7528Chroma<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-56\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E9%80%9A%E8%BF%87Retriever%E6%A3%80%E7%B4%A2%E5%99%A8%E7%9A%84%E6%96%B9%E5%BC%8F%E4%BD%BF%E7%94%A8Chroma\" >\u901a\u8fc7Retriever\u68c0\u7d22\u5668\u7684\u65b9\u5f0f\u4f7f\u7528Chroma<\/a><ul class='ez-toc-list-level-5' ><li class='ez-toc-heading-level-5'><a class=\"ez-toc-link ez-toc-heading-57\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E4%BD%BF%E7%94%A8%E6%96%B9%E6%B3%95\" >\u4f7f\u7528\u65b9\u6cd5<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-5'><a class=\"ez-toc-link ez-toc-heading-58\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E%EF%BC%9A\" >\u51fd\u6570\u8bf4\u660e\uff1a<\/a><\/li><\/ul><\/li><\/ul><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-59\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\" >\u5185\u5bb9\u5c0f\u7ed3<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-60\" href=\"https:\/\/17aitech.com\/?p=18186\/#%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\" >\u53c2\u8003\u8d44\u6599<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"%E5%89%8D%E8%A8%80\"><\/span>\u524d\u8a00<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>\u5728\u4e0a\u4e00\u7ae0<a href=\"https:\/\/17aitech.com\/?p=15918\">\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day27\uff1a\u5927\u6a21\u578b\u4e4b\u521d\u8bc6RAG<\/a>\u4e2d\uff0c\u6211\u4eec\u521d\u6b65\u4e86\u89e3\u4e86RAG\u7684\u57fa\u672c\u6982\u5ff5\u548c\u539f\u7406\uff0c\u5e76\u901a\u8fc7\u4ee3\u7801\u5b9e\u8df5\u4e86\u4e00\u4e2a\u7b80\u5355\u7684RAG\u6d41\u7a0b\u3002\u672c\u7ae0\u6211\u4eec\u5c06\u57fa\u4e8eRAG\u7684\u57fa\u672c\u6d41\u7a0b\uff0c\u6df1\u5165\u4e86\u89e3<code>\u6587\u6863\u8bfb\u53d6(LOAD)<\/code>\u3001<code>\u6587\u6863\u5207\u5206(SPLIT)<\/code>\u3001<code>\u5411\u91cf\u5316(EMBED)<\/code> \u548c <code>\u5b58\u50a8(STORE)<\/code> \u7684\u6bcf\u4e2a\u73af\u8282\uff0c\u5e76\u7ed3\u5408\u4ee3\u7801\u8fdb\u884c\u5e38\u89c1\u573a\u666f\u7684\u5b9e\u8df5\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"RAG%E6%B5%81%E7%A8%8B%E5%9B%9E%E9%A1%BE\"><\/span>RAG\u6d41\u7a0b\u56de\u987e<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png\" alt=\"\" \/><\/a><\/p>\n<p>\u56de\u987eRAG\u7684\u6d41\u7a0b\u5982\u4e0a\u6240\u793a\uff0c\u5177\u4f53\u4ee3\u7801\u89c1<a href=\"https:\/\/17aitech.com\/?p=15918#toc-9\">RAG\u4ee3\u7801<\/a>\uff0c\u672c\u7ae0\u4e0d\u518d\u8d58\u8ff0\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"%E6%96%87%E6%A1%A3%E8%AF%BB%E5%8F%96LOAD\"><\/span>\u6587\u6863\u8bfb\u53d6(LOAD)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><strong>\u7b80\u4ecb<\/strong>\uff1a\u7531\u4e8e\u6211\u4eec\u7684\u77e5\u8bc6\u5e7f\u6cdb\u5b58\u5728\u5404\u7c7b\u6587\u6863\u4e2d\uff0c\u6240\u4ee5\u6587\u6863\u8bfb\u53d6\uff08<code>LOAD<\/code>\uff09\u5728RAG\u7cfb\u7edf\u4e2d\u627f\u62c5\u7740 <code>\u83b7\u53d6\u6570\u636e<\/code> \u7684\u804c\u8d23\u3002<\/p>\n<p><strong>\u5e38\u89c1\u6587\u4ef6<\/strong>\uff1a<\/p>\n<ul>\n<li>\u6587\u672c\u6587\u4ef6\uff08.txt\uff09<\/li>\n<li>CSV\uff08.csv\uff09<\/li>\n<li>PDF\uff08.pdf\uff09<\/li>\n<li>Word \u6587\u6863\uff08.docx\uff09<\/li>\n<li>Excel \u5de5\u4f5c\u7c3f\uff08.xlsx\uff09<\/li>\n<li>PPT \u6f14\u793a\u6587\u7a3f\uff08.pptx\uff09<\/li>\n<li>HTML \u6587\u4ef6\uff08.html\uff09<\/li>\n<li>Markdown \u6587\u4ef6\uff08.md\uff09<\/li>\n<\/ul>\n<p><strong>\u5e38\u89c1\u7684\u6587\u4ef6\u5904\u7406\u5e93<\/strong>\uff1a<\/p>\n<ul>\n<li>langchain_community.document_loaders<\/li>\n<\/ul>\n<blockquote>\n<ul>\n<li>\u5728langchain_community\u7684<a href=\"https:\/\/api.python.langchain.com\/en\/latest\/community_api_reference.html#module-langchain_community.document_loaders\">API\u6587\u6863<\/a>\u4e2d\u6709\u6570\u5341\u79cd\u6587\u4ef6\u7684\u5904\u7406API\uff0c\u9664\u4e86\u4e0a\u8ff0\u5e38\u89c1\u7684\u6587\u4ef6\u4e4b\u5916\uff0c\u8fd8\u6709.srt\u5b57\u5e55\u6587\u4ef6\u3001.ipynb\u6587\u4ef6\u3001\u4ee3\u7801\u7247\u6bb5\u7b49\u652f\u6301\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<h3><span class=\"ez-toc-section\" id=\"CSV%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>CSV\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"CSVLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>CSVLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.csv_loader.CSVLoader(\n    file_path: Union[str, Path], \n    source_column: Optional[str] = None, \n    metadata_columns: Sequence[str] = (), \n    csv_args: Optional[Dict] = None, \n    encoding: Optional[str] = None, \n    autodetect_encoding: bool = False)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li><code>file_path<\/code>:\u6307\u5b9a\u8981\u52a0\u8f7d\u7684 CSV \u6587\u4ef6\u7684\u8def\u5f84\n<ul>\n<li>\u4f8b\u5982\uff1a&quot;data\/myfile.csv&quot;<\/li>\n<\/ul>\n<\/li>\n<li><code>source_column<\/code>(\u53ef\u9009):\u6307\u5b9a\u54ea\u4e00\u5217\u4f5c\u4e3a\u6570\u636e\u7684\u4e3b\u8981\u6765\u6e90\u3002\n<ul>\n<li>\u4f8b\u5982\uff1a&quot;text&quot;<\/li>\n<\/ul>\n<\/li>\n<li><code>metadata_columns<\/code>:\u6307\u5b9a\u54ea\u4e9b\u5217\u5e94\u4f5c\u4e3a\u5143\u6570\u636e\u52a0\u8f7d\u3002\n<ul>\n<li>\u4f8b\u5982\uff1a[&quot;name&quot;, &quot;age&quot;]<\/li>\n<\/ul>\n<\/li>\n<li><code>csv_args<\/code>(\u53ef\u9009):\u81ea\u5b9a\u4e49 CSV \u52a0\u8f7d\u7684\u884c\u4e3a\u3002\n<ul>\n<li>\u4f8b\u5982\uff1a{&quot;delimiter&quot;: &quot;;&quot;, &quot;header&quot;: 0}\uff08\u6307\u5b9a\u5206\u9694\u7b26\u4e3a\u5206\u53f7\uff0c\u5e76\u5c06\u7b2c\u4e00\u884c\u4f5c\u4e3a\u8868\u5934\uff09<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p><strong>\u793a\u4f8b1<\/strong>\uff1a\u4ec5\u52a0\u8f7d.csv\u6587\u4ef6\u793a\u4f8b\uff1a<\/p>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import CSVLoader\n\nstudents_loader = CSVLoader(file_path=&quot;testfiles\/students.csv&quot;)\ndocs = students_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-python\"># [Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 0}, page_content=&#039;name: Tom\\nage: 12\\nscore: 77&#039;),\n#  Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 1}, page_content=&#039;name: Jerry\\nage: 11\\nscore: 88&#039;),\n#  Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 2}, page_content=&#039;name: Jim\\nage: 12\\nscore: 96&#039;)]<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>metadata\uff1a\u6587\u6863\u5143\u6570\u636e\uff0c\u5305\u62ec\u6587\u6863\u6765\u6e90\u3001\u884c\u53f7\u7b49\uff0c\u7528\u6765\u8fdb\u884c\u6eaf\u6e90\u4f7f\u7528\u3002<\/li>\n<li>page_content\uff1a\u6587\u6863\u5185\u5bb9\uff0c\u5373\u6587\u672c\u5185\u5bb9\u3002<\/li>\n<\/ul>\n<p><strong>\u793a\u4f8b2<\/strong>\uff1a\u52a0\u8f7d.csv\u6587\u4ef6\u6307\u5b9a\u5217\u4f5c\u4e3a\u5143\u6570\u636e\uff1a<\/p>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import CSVLoader\n\nstudents_loader = CSVLoader(file_path=&quot;testfiles\/students.csv&quot;, metadata_columns=[&quot;name&quot;, &quot;score&quot;])\ndocs = students_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-python\"># [Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 0, &#039;name&#039;: &#039;Tom&#039;, &#039;score&#039;: &#039;77&#039;}, page_content=&#039;age: 12&#039;),\n#  Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 1, &#039;name&#039;: &#039;Jerry&#039;, &#039;score&#039;: &#039;88&#039;}, page_content=&#039;age: 11&#039;),\n#  Document(metadata={&#039;source&#039;: &#039;testfiles\/students.csv&#039;, &#039;row&#039;: 2, &#039;name&#039;: &#039;Jim&#039;, &#039;score&#039;: &#039;96&#039;}, page_content=&#039;age: 12&#039;)]<\/code><\/pre>\n<p><strong>\u793a\u4f8b3<\/strong>\uff1a\u4f7f\u7528RAG\u94fe\u6d4b\u8bd5\u5927\u6a21\u578b\u7684\u77e5\u8bc6\u7406\u89e3\u80fd\u529b<\/p>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import CSVLoader\nfrom langchain_core.output_parsers import StrOutputParser\nfrom langchain_core.runnables import RunnablePassthrough\nfrom langchain_chroma import Chroma\nfrom langchain_core.prompts import ChatPromptTemplate\nfrom utils import get_qwen_models\n\ndef format_docs(docs):\n    return &quot;\\n\\n&quot;.join(doc.page_content for doc in docs)\n\n# \u8fde\u63a5\u5927\u6a21\u578b\nllm, chat, embed = get_qwen_models()\n\n# \u8bfb\u53d6.csv\u6587\u4ef6\nstudents_loader = CSVLoader(file_path=&quot;testfiles\/students.csv&quot;)\ndocs = students_loader.load()\n\n# \u5411\u91cf\u5316\u5165\u5e93\nvectorstore = Chroma.from_documents(documents=docs, embedding=embed)\n\nretriever = vectorstore.as_retriever(search_kwargs={&quot;k&quot;: 2})\n\nprompt = ChatPromptTemplate.from_messages([\n  (&quot;human&quot;, &quot;&quot;&quot;You are an assistant for question-answering tasks. \nUse the following pieces of retrieved context to answer the question. \nIf you don&#039;t know the answer, just say that you don&#039;t know. \nUse three sentences maximum and keep the answer concise.\nQuestion: {question} \nContext: {context} \nAnswer:&quot;&quot;&quot;)\n])\n\n# RAG \u94fe\nrag_chain = (\n    {&quot;context&quot;: retriever | format_docs, \n     &quot;question&quot;: RunnablePassthrough()}\n    | prompt\n    | chat\n    | StrOutputParser()\n)\n\nrag_chain.invoke(input=&quot;\u8c01\u7684\u6210\u7ee9\u6700\u9ad8\uff1f&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-python\"># &#039;Jerry\u7684\u6210\u7ee9\u6700\u9ad8\uff0c\u5f97\u4e8688\u5206\u3002&#039;<\/code><\/pre>\n<h3><span class=\"ez-toc-section\" id=\"TXT%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>TXT\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"TXTLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>TXTLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.text.TextLoader(\n    file_path: Union[str, Path], \n    encoding: Optional[str] = None, \n    autodetect_encoding: bool = False)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li><code>file_path<\/code>:\u6307\u5b9a\u8981\u52a0\u8f7d\u7684\u6587\u672c\u6587\u4ef6\u7684\u8def\u5f84\u3002<\/li>\n<li><code>encoding<\/code>(\u53ef\u9009):\u6307\u5b9a\u6587\u672c\u6587\u4ef6\u7684\u7f16\u7801\u683c\u5f0f\uff0c\u4e00\u822c\u662f&quot;utf-8&quot;<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-2\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\n\ntxt_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\n\ndocs = txt_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">[Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\\n\\n1. \u673a\u5668\u7ffb\u8bd1\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\\n\\n2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\u53ca\u65f6\u53d1\u73b0\u95ee\u9898\u5e76\u4f18\u5316\u670d\u52a1\u3002\\n\\n3. \u6587\u672c\u751f\u6210\u4e0e\u521b\u4f5c\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u751f\u6210\u7b26\u5408\u8bed\u6cd5\u89c4\u5219\u7684\u6587\u7ae0\u3001\u65b0\u95fb\u3001\u5c0f\u8bf4\u7b49\u6587\u672c\u5185\u5bb9\u3002\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u6587\u672c\u6570\u636e\uff0c\u6a21\u578b\u53ef\u4ee5\u751f\u6210\u5177\u6709\u521b\u9020\u6027\u548c\u76f8\u5173\u6027\u7684\u5185\u5bb9\uff0c\u5e7f\u6cdb\u5e94\u7528\u4e8e\u65b0\u95fb\u62a5\u9053\u3001\u5e7f\u544a\u8425\u9500\u7b49\u9886\u57df\u3002\u6b64\u5916\uff0c\u5b83\u8fd8\u80fd\u6839\u636e\u7ed9\u5b9a\u4e3b\u9898\u6216\u5173\u952e\u8bcd\u751f\u6210\u6587\u7ae0\uff0c\u4e3a\u521b\u4f5c\u8005\u63d0\u4f9b\u7075\u611f\u548c\u8f85\u52a9\u3002\\n\\n4. \u60c5\u611f\u5206\u6790\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u5206\u6790\u6587\u672c\u4e2d\u7684\u60c5\u611f\u503e\u5411\u548c\u60c5\u611f\u8868\u8fbe\uff0c\u5e2e\u52a9\u4f01\u4e1a\u4e86\u89e3\u5ba2\u6237\u53cd\u9988\u548c\u60c5\u611f\u72b6\u6001\uff0c\u4ece\u800c\u5236\u5b9a\u66f4\u7cbe\u51c6\u7684\u8425\u9500\u7b56\u7565\u6216\u4f18\u5316\u5ba2\u6237\u670d\u52a1\u3002\u8fd9\u79cd\u6280\u672f\u8fd8\u53ef\u7528\u4e8e\u793e\u4ea4\u5a92\u4f53\u76d1\u63a7\uff0c\u5b9e\u65f6\u5206\u6790\u516c\u4f17\u5bf9\u67d0\u4e00\u4e3b\u9898\u6216\u4e8b\u4ef6\u7684\u60c5\u7eea\u548c\u53cd\u5e94\u3002\\n\\n5. \u81ea\u52a8\u95ee\u7b54\u7cfb\u7edf\\n\u63cf\u8ff0\uff1a\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u95ee\u9898\u548c\u7b54\u6848\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u81ea\u52a8\u751f\u6210\u7b26\u5408\u8bed\u6cd5\u89c4\u5219\u7684\u95ee\u9898\u548c\u7b54\u6848\u3002\u8fd9\u79cd\u81ea\u52a8\u95ee\u7b54\u7cfb\u7edf\u53ef\u5e94\u7528\u4e8e\u667a\u80fd\u52a9\u624b\u3001\u641c\u7d22\u5f15\u64ce\u7b49\u9886\u57df\uff0c\u4e3a\u7528\u6237\u63d0\u4f9b\u9ad8\u6548\u3001\u51c6\u786e\u7684\u4fe1\u606f\u670d\u52a1\u3002\u7ed3\u5408\u77e5\u8bc6\u56fe\u8c31\u6280\u672f\uff0c\u95ee\u7b54\u7cfb\u7edf\u7684\u77e5\u8bc6\u68c0\u7d22\u548c\u63a8\u7406\u80fd\u529b\u5f97\u5230\u8fdb\u4e00\u6b65\u589e\u5f3a\u3002\\n\\n6. \u81ea\u52a8\u6458\u8981\u4e0e\u603b\u7ed3\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u81ea\u52a8\u5bf9\u6587\u672c\u8fdb\u884c\u6458\u8981\u548c\u603b\u7ed3\uff0c\u63d0\u53d6\u5173\u952e\u4fe1\u606f\uff0c\u5e2e\u52a9\u7528\u6237\u5feb\u901f\u4e86\u89e3\u6587\u672c\u4e3b\u65e8\u548c\u91cd\u70b9\u3002\u8fd9\u79cd\u6280\u672f\u5728\u5b66\u672f\u8bba\u6587\u3001\u65b0\u95fb\u62a5\u9053\u7b49\u9886\u57df\u5177\u6709\u91cd\u8981\u5e94\u7528\u4ef7\u503c\uff0c\u63d0\u9ad8\u4e86\u4fe1\u606f\u83b7\u53d6\u7684\u6548\u7387\u3002\\n\\n7. \u4ee3\u7801\u751f\u6210\u4e0e\u81ea\u52a8\u5316\u7f16\u7a0b\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u4ee3\u7801\u6570\u636e\uff0c\u53ef\u4ee5\u7406\u89e3\u7f16\u7a0b\u8bed\u8a00\u7684\u8bed\u6cd5\u548c\u903b\u8f91\u89c4\u5219\uff0c\u5b9e\u73b0\u4ee3\u7801\u7684\u81ea\u52a8\u751f\u6210\u3002\u8fd9\u6709\u52a9\u4e8e\u975e\u6280\u672f\u7528\u6237\u751f\u6210\u57fa\u672c\u4ee3\u7801\uff0c\u540c\u65f6\u4e3a\u4e13\u4e1a\u7f16\u7a0b\u4eba\u5458\u63d0\u4f9b\u8f85\u52a9\uff0c\u52a0\u5feb\u5f00\u53d1\u8fdb\u7a0b\u3002\u7136\u800c\uff0c\u5728\u590d\u6742\u4efb\u52a1\u4e2d\u4ecd\u9700\u4eba\u5de5\u5ba1\u6838\u548c\u8c03\u6574\u3002\\n\\n8. \u4fe1\u606f\u68c0\u7d22\u4e0e\u63a8\u8350\u7cfb\u7edf\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u53ef\u5e94\u7528\u4e8e\u6539\u5584\u641c\u7d22\u5f15\u64ce\u7ed3\u679c\u548c\u5185\u5bb9\u63a8\u8350\u7b97\u6cd5\u3002\u901a\u8fc7\u5206\u6790\u7528\u6237\u67e5\u8be2\u610f\u56fe\u548c\u4e0a\u4e0b\u6587\u4fe1\u606f\uff0c\u6a21\u578b\u80fd\u591f\u63d0\u4f9b\u66f4\u51c6\u786e\u3001\u4e2a\u6027\u5316\u7684\u641c\u7d22\u7ed3\u679c\u548c\u5185\u5bb9\u63a8\u8350\uff0c\u63d0\u5347\u7528\u6237\u4f53\u9a8c\u548c\u6ee1\u610f\u5ea6\u3002\\n\\n9. \u751f\u7269\u533b\u5b66\u7814\u7a76\\n\u63cf\u8ff0\uff1a\u5728\u751f\u7269\u533b\u5b66\u9886\u57df\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u53ef\u7528\u4e8e\u5206\u6790\u57fa\u56e0\u7ec4\u6570\u636e\u3001\u86cb\u767d\u8d28\u76f8\u4e92\u4f5c\u7528\u7b49\uff0c\u52a0\u901f\u836f\u7269\u53d1\u73b0\u548c\u65b0\u7597\u6cd5\u7684\u7814\u7a76\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u9884\u6d4b\u57fa\u56e0\u53d8\u5f02\u7684\u529f\u80fd\u5f71\u54cd\uff0c\u7814\u7a76\u8005\u80fd\u591f\u66f4\u5168\u9762\u5730\u5206\u6790\u4eba\u7c7b\u57fa\u56e0\u7ec4\u7684\u6f5c\u5728\u98ce\u9669\u548c\u6cbb\u7597\u9776\u70b9\u3002\\n\\n10. \u8bed\u97f3\u8bc6\u522b\u4e0e\u751f\u6210\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u5728\u8bed\u97f3\u8bc6\u522b\u548c\u8bed\u97f3\u751f\u6210\u65b9\u9762\u4e5f\u5c55\u73b0\u51fa\u5de8\u5927\u6f5c\u529b\u3002\u901a\u8fc7\u5c06\u8bed\u97f3\u8f6c\u5f55\u4e3a\u6587\u672c\u6216\u5c06\u6587\u672c\u8f6c\u5316\u4e3a\u8bed\u97f3\uff0c\u8be5\u6280\u672f\u4f7f\u5f97\u4eba\u4eec\u4e0e\u8ba1\u7b97\u673a\u7684\u4ea4\u4e92\u66f4\u52a0\u81ea\u7136\u548c\u4fbf\u6377\u3002\u8fd9\u5bf9\u4e8e\u6709\u542c\u529b\u6216\u89c6\u89c9\u969c\u788d\u7684\u4eba\u7fa4\u5c24\u4e3a\u91cd\u8981\uff0c\u6709\u52a9\u4e8e\u4ed6\u4eec\u66f4\u597d\u5730\u7406\u89e3\u548c\u4eab\u53d7\u97f3\u89c6\u9891\u5185\u5bb9\u3002\\n\\n\u7efc\u4e0a\u6240\u8ff0\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u5728\u591a\u4e2a\u9886\u57df\u90fd\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u524d\u666f\u548c\u5de8\u5927\u7684\u4ef7\u503c\u6f5c\u529b\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\u548c\u5b8c\u5584\uff0c\u6211\u4eec\u6709\u7406\u7531\u76f8\u4fe1\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u5728\u672a\u6765\u7684\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d1\u6325\u66f4\u52a0\u91cd\u8981\u7684\u4f5c\u7528\u3002&#039;)]<\/code><\/pre>\n<h3><span class=\"ez-toc-section\" id=\"PDF%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>PDF\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"PyMuPDFLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>PyMuPDFLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.pdf.PyMuPDFLoader(\n    file_path: str, *, \n    headers: Optional[Dict] = None, \n    extract_images: bool = False, \n    **kwargs: Any)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li><code>extract_images<\/code>\uff1a\u6307\u793a\u662f\u5426\u4ece PDF \u4e2d\u63d0\u53d6\u56fe\u50cf\u3002\u5982\u679c\u8bbe\u7f6e\u4e3a True\uff0c\u5c06\u5c1d\u8bd5\u63d0\u53d6 PDF \u4e2d\u7684\u6240\u6709\u56fe\u50cf\u3002<\/li>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install PyMuPDF<\/code><\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-3\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import PyMuPDFLoader\n\npdf_loader = PyMuPDFLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.pdf&quot;)\n\ndocs = pdf_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PDF\u4ee3\u7801\u52a0\u8f7d\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PDF\u4ee3\u7801\u52a0\u8f7d\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<pre><code class=\"language-bash\"># metadata\u6570\u636e\u8be6\u60c5\nmetadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.pdf&#039;, \n&#039;file_path&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.pdf&#039;, \n&#039;page&#039;: 0, &#039;total_pages&#039;: 21, \n&#039;format&#039;: &#039;PDF 1.6&#039;, \n&#039;title&#039;: &#039;JSON \u662f\u4ec0\u4e48\uff1f&#039;, \n&#039;author&#039;: &#039;\u756a\u8304\u82b1\u56ed&#039;, \n&#039;subject&#039;: &#039;&#039;, \n&#039;keywords&#039;: &#039;&#039;, \n&#039;creator&#039;: &#039;WPS Office \u4e2a\u4eba\u7248&#039;, \n&#039;producer&#039;: &#039;PDFlib 7.0.3 (C++\/Win32)&#039;, \n&#039;creationDate&#039;: &quot;D:20090304181435+08&#039;00&#039;&quot;, \n&#039;modDate&#039;: &quot;D:20100525102623+08&#039;00&#039;&quot;, &#039;trapped&#039;: &#039;&#039;\n}<\/code><\/pre>\n<p>\u539f\u59cbPDF\u6587\u4ef6\u5982\u4e0b\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PDF\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PDF\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u8bf4\u660e\uff1a\u901a\u8fc7\u5bf9\u6bd4\u53ef\u4ee5\u770b\u5230<\/p>\n<ul>\n<li>pagecontent\u4e0ePDF\u6587\u4ef6\u5185\u5bb9\u4e00\u81f4\u3002<\/li>\n<li>metadata\u6570\u636e\u6709\u6bd4\u8f83\u591a\u7684\u5b57\u6bb5\uff0c\u4f8b\u5982\uff1aformat\uff0ctitle\uff0cauthor\uff0csubject\uff0ckeywords\uff0ccreator\uff0cproducer\uff0ccreationDate\uff0cmodDate\u7b49\u3002<\/li>\n<\/ul>\n<h3><span class=\"ez-toc-section\" id=\"EXCEL%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>EXCEL\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"UnstructuredExcelLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>UnstructuredExcelLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.excel.UnstructuredExcelLoader(\n    file_path: Union[str, Path], \n    mode: str = &#039;single&#039;, \n    **unstructured_kwargs: Any)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install unstructured<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-4\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import UnstructuredExcelLoader\n\nexcel_loader = UnstructuredExcelLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.xlsx&quot;, encoding=&quot;utf8&quot;)\n\ndocs = excel_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/EXCEL\u4ee3\u7801\u52a0\u8f7d\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/EXCEL\u4ee3\u7801\u52a0\u8f7d\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/EXCEL\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/EXCEL\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"PPT%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>PPT\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"UnstructuredPowerPointLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>UnstructuredPowerPointLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.powerpoint.UnstructuredPowerPointLoader(\n    file_path: Union[str, List[str], Path, List[Path]], *, \n    mode: str = &#039;single&#039;, \n    **unstructured_kwargs: Any)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install python-pptx<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-5\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import UnstructuredPowerPointLoader\n\nppt_loader = UnstructuredPowerPointLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.pptx&quot;, mode=&quot;single&quot;)\n\ndocs = ppt_loader.load()\n\ndocs[0].page_content<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PPT\u4ee3\u7801\u8fd0\u884c\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PPT\u4ee3\u7801\u8fd0\u884c\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PPT\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/PPT\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u901a\u8fc7\u5bf9\u6bd4\uff0c\u53ef\u4ee5\u770b\u5230PPT\u4e2d\u7684\u6587\u5b57\u4f1a\u88ab\u63d0\u53d6\u51fa\u6765\uff0c\u56fe\u7247\u4f1a\u88ab\u6e05\u6d17\u6389\u3002<\/li>\n<\/ul>\n<h3><span class=\"ez-toc-section\" id=\"WORD%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>WORD\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"UnstructuredWordDocumentLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>UnstructuredWordDocumentLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.word_document.UnstructuredWordDocumentLoader(\n    file_path: Union[str, List[str], Path, List[Path]], *, \n    mode: str = &#039;single&#039;, \n    **unstructured_kwargs: Any)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install python-docx<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-6\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import UnstructuredWordDocumentLoader\n\nword_loader = UnstructuredWordDocumentLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.docx&quot;, mode=&quot;single&quot;)\n\ndocs = ppt_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/word\u4ee3\u7801\u6267\u884c\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/word\u4ee3\u7801\u6267\u884c\u622a\u56fe.png\" alt=\"\" \/><\/a><br \/>\n\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/word\u539f\u59cb\u6587\u4ef6.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/word\u539f\u59cb\u6587\u4ef6.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"MARKDOWN%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>MARKDOWN\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"UnstructuredMarkdownLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>UnstructuredMarkdownLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.markdown.UnstructuredMarkdownLoader(\n    file_path: Union[str, List[str], Path, List[Path]], *, \n    mode: str = &#039;single&#039;, \n    **unstructured_kwargs: Any)<\/code><\/pre>\n<ul>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install markdown<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-7\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import UnstructuredMarkdownLoader\n\nmd_loader = UnstructuredMarkdownLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.md&quot;, mode=&quot;single&quot;)\n\ndocs = md_loader.load()\n\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/markdown\u6587\u4ef6\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/markdown\u6587\u4ef6\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/markdown\u539f\u59cb\u6587\u4ef6.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/markdown\u539f\u59cb\u6587\u4ef6.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"HTML%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>HTML\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"UnstructuredHTMLLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>UnstructuredHTMLLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.html.UnstructuredHTMLLoader(\n    file_path: Union[str, List[str], Path, List[Path]], *, \n    mode: str = &#039;single&#039;, \n    **unstructured_kwargs: Any)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>model\u53ef\u4ee5\u8bbe\u7f6e\u4e3a&#8217;elements&#8217;\uff0c\u8868\u793a\u5c06HTML\u5143\u7d20\u4f5c\u4e3a\u6587\u6863\u8fdb\u884c\u5207\u5206\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-8\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import UnstructuredHTMLLoader\n\nhtml_file = &quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.htm&quot;\n\nhtml_loader = UnstructuredHTMLLoader(file_path=html_file, mode=&quot;single&quot;)\n\ndocs = html_loader.load()\n\ndocs[0].page_content<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u6587\u4ef6\u8fd0\u884c\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u6587\u4ef6\u8fd0\u884c\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u539f\u59cb\u6587\u4ef6\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u8865\u5145\u8bf4\u660e\uff1a<br \/>\n\u5982\u679c\u5c06\u4e0a\u9762\u7684mode\u6539\u4e3amode=&quot;elements&quot;,\u8fd0\u884c\u5982\u4e0b\uff0c\u4f46\u662f\u5f97\u5230\u5185\u5bb9\u5bf9\u4e8e\u5efa\u7acb\u77e5\u8bc6\u5e93\u6ca1\u6709\u592a\u5927\u5e2e\u52a9\u3002<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u6587\u4ef6element\u622a\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/html\u6587\u4ef6element\u622a\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"SQL%E6%96%87%E4%BB%B6%E5%8A%A0%E8%BD%BD%E5%99%A8\"><\/span>SQL\u6587\u4ef6\u52a0\u8f7d\u5668<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"SQLDatabaseLoader%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>SQLDatabaseLoader\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_community.document_loaders.sql_database.SQLDatabaseLoader(\n    query: Union[str, Select], \n    db: SQLDatabase, *, \n    parameters: Optional[Dict[str, Any]] = None, \n    page_content_mapper: Optional[Callable[[...], str]] = None, \n    metadata_mapper: Optional[Callable[[...], Dict[str, Any]]] = None, \n    source_columns: Optional[Sequence[str]] = None, \n    include_rownum_into_metadata: bool = False, \n    include_query_into_metadata: bool = False)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li><code>query<\/code>\uff1a\u6307\u5b9a\u8981\u6267\u884c\u7684 SQL \u67e5\u8be2\u3002\n<ul>\n<li>\u4f8b\u5982\uff1aquery = &quot;SELECT * FROM documents&quot;<\/li>\n<\/ul>\n<\/li>\n<li><code>db<\/code>: \u6307\u5b9a\u8981\u8fde\u63a5\u7684 SQL \u6570\u636e\u5e93\u5b9e\u4f8b\u3002<\/li>\n<li><code>parameters<\/code>(\u53ef\u9009)\uff1a\u53ef\u9009\u53c2\u6570\uff0c\u7528\u4e8e\u4f20\u9012 SQL \u67e5\u8be2\u7684\u53c2\u6570\uff0c\u4ee5\u4fbf\u4e8e\u52a8\u6001\u67e5\u8be2\u3002\n<ul>\n<li>\u4f8b\u5982\uff1aparameters = {&quot;id&quot;: 1}<\/li>\n<\/ul>\n<\/li>\n<li><code>page_content_mapper<\/code>(\u53ef\u9009)\uff1a\u53ef\u9009\u7684\u6620\u5c04\u51fd\u6570\uff0c\u7528\u4e8e\u5904\u7406\u67e5\u8be2\u7ed3\u679c\u4e2d\u7684\u6bcf\u4e00\u884c\uff0c\u8fd4\u56de\u683c\u5f0f\u5316\u540e\u7684\u5185\u5bb9\u3002<\/li>\n<li><code>metadata_mapper<\/code>(\u53ef\u9009)\uff1a\u53ef\u9009\u7684\u6620\u5c04\u51fd\u6570\uff0c\u7528\u4e8e\u5904\u7406\u67e5\u8be2\u7ed3\u679c\u4e2d\u7684\u6bcf\u4e00\u884c\uff0c\u8fd4\u56de\u5143\u6570\u636e\u5b57\u5178\u3002\n<ul>\n<li>\u4f8b\u5982\uff1ametadata_mapper=lambda row: {&quot;id&quot;: row[&#8216;id&#8217;], &quot;created_at&quot;: row[&#8216;created_at&#8217;]}<\/li>\n<\/ul>\n<\/li>\n<li><code>source_columns<\/code>(\u53ef\u9009)\uff1a\u53ef\u9009\u53c2\u6570\uff0c\u6307\u5b9a\u8981\u4ece\u67e5\u8be2\u7ed3\u679c\u4e2d\u63d0\u53d6\u7684\u5217\u540d\u3002<\/li>\n<li><code>include_rownum_into_metadata<\/code>\uff1a\u6307\u793a\u662f\u5426\u5c06\u884c\u53f7\u5305\u542b\u5728\u5143\u6570\u636e\u4e2d\u3002<\/li>\n<li><code>include_query_into_metadata<\/code>\uff1a\u6307\u793a\u662f\u5426\u5c06\u6267\u884c\u7684 SQL \u67e5\u8be2\u5305\u542b\u5728\u5143\u6570\u636e\u4e2d\u3002<\/li>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\uff1a<code>pip install sqlalchemy<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-9\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import SQLDatabaseLoader\nfrom langchain_community.utilities import SQLDatabase\nfrom sqlalchemy import create_engine\n\n# \u521b\u5efa SQLite \u6570\u636e\u5e93\u5f15\u64ce\uff0c\u6307\u5b9a\u6570\u636e\u5e93\u6587\u4ef6\u7684\u8def\u5f84\nengine = create_engine(url=&quot;sqlite:\/\/\/testfiles\/students.db&quot;)\n\n# \u4f7f\u7528\u521b\u5efa\u7684\u5f15\u64ce\u5b9e\u4f8b\u5316 SQLDatabase \u5bf9\u8c61\ndb = SQLDatabase(engine=engine)\n\n# \u521b\u5efa SQLDatabaseLoader \u5b9e\u4f8b\uff0c\u6307\u5b9a\u8981\u6267\u884c\u7684 SQL \u67e5\u8be2\nsql_loader = SQLDatabaseLoader(query=&quot;SELECT * FROM stu_score;&quot;, db=db, \n                               include_query_into_metadata=True,\n                               include_rownum_into_metadata=True)\n\ndocs = sql_loader.load()\ndocs<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/SQL\u6267\u884c\u7ed3\u679c.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/SQL\u6267\u884c\u7ed3\u679c.png\" alt=\"\" \/><\/a><br \/>\n\u539f\u59cb\u6587\u4ef6\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/SQL\u539f\u59cb\u6587\u4ef6.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/SQL\u539f\u59cb\u6587\u4ef6.png\" alt=\"\" \/><\/a><\/p>\n<p>\u4ee5\u4e0a\u53ea\u662f\u5bf9\u5404\u7c7b\u6587\u4ef6\u52a0\u8f7d\u5668\u7684\u7b80\u5355\u68b3\u7406\uff0c\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u9700\u8981\u8fdb\u884c\u8fdb\u4e00\u6b65\u7684\u5c01\u88c5\uff0c\u4f8b\u5982\uff1a<\/p>\n<ul>\n<li>\u5c01\u88c5\u4e3aclass\u65b9\u5f0f<\/li>\n<li>\u4ee5\u6587\u4ef6\u540e\u7f00\u540d\u81ea\u52a8\u5224\u65ad\u8c03\u7528\u7684\u52a0\u8f7d\u5668<\/li>\n<li>\u6dfb\u52a0\u66f4\u591a\u56de\u6eaf\u4fe1\u606f<\/li>\n<li>&#8230;&#8230;<\/li>\n<\/ul>\n<blockquote>\n<p>\u5728\u5f00\u6e90\u9879\u76eeQAnything(\u7f51\u6613\u7684\u4e00\u6b3e\u5f00\u6e90RAG\u7cfb\u7edf)\u4e2d\uff0c\u8be5\u9879\u76ee\u4e2d\u7684<a href=\"https:\/\/github.com\/netease-youdao\/QAnything\/blob\/master\/qanything_kernel\/core\/local_file.py\">qanything_kernel<\/a>\u6709\u7740\u6bd4\u8f83\u8be6\u7ec6\u7684\u5c01\u88c5\uff0c\u53ef\u4ee5\u53c2\u8003\u3002<\/p>\n<\/blockquote>\n<h2><span class=\"ez-toc-section\" id=\"%E6%96%87%E6%A1%A3%E5%88%87%E5%88%86SPLIT\"><\/span>\u6587\u6863\u5207\u5206(SPLIT)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3><span class=\"ez-toc-section\" id=\"%E6%8C%89%E7%85%A7%E5%AD%97%E7%AC%A6%E5%88%87%E5%88%86\"><\/span>\u6309\u7167\u5b57\u7b26\u5207\u5206<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"CharacterTextSplitter%E7%B1%BB%E8%AF%B4%E6%98%8E\"><\/span>CharacterTextSplitter\u7c7b\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_text_splitters.character.CharacterTextSplitter:\n    def __init__(self, chunk_size: int, chunk_overlap: int = 0):<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>chunk_size\uff1a\u6307\u5b9a\u6bcf\u4e2a\u6587\u672c\u7247\u6bb5\u7684\u6700\u5927\u5b57\u7b26\u957f\u5ea6\u3002<\/li>\n<li>chunk_overlap\uff1a\u6307\u5b9a\u76f8\u90bb\u6587\u672c\u7247\u6bb5\u4e4b\u95f4\u7684\u91cd\u53e0\u5b57\u7b26\u6570\u3002\u91cd\u53e0\u53ef\u4ee5\u5e2e\u52a9\u4fdd\u6301\u4e0a\u4e0b\u6587\u7684\u4e00\u81f4\u6027\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-10\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import CharacterTextSplitter\n\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\n\ndocs = text_loader.load()\n\n# \u5b57\u7b26\u7ea7\u6587\u672c\u5207\u5206\u5668\nspliter = CharacterTextSplitter(separator = &#039;\\n\\n&#039;, chunk_size=128, chunk_overlap=32)\n\nresults = spliter.split_documents(documents=docs)\n\n# \u6253\u5370\u5207\u5206\u7ed3\u679c\nprint(f&#039;\u5207\u5206\u4e2a\u6570\uff1a{len(results)}&#039;)\n\nfor idx, chunk in enumerate(results):\n    print(f&#039;\u7b2c{idx+1}\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a{len(chunk.page_content)}&#039;)\n    print(f&#039;\u5207\u5206\u5185\u5bb9\uff1a\\n{chunk.page_content}&#039;)\n    print(&quot;-*-&quot; *30 + &quot;\\n&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\"># chunk_size=128, chunk_overlap=32\n\u5207\u5206\u4e2a\u6570\uff1a12\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a50\n\u5207\u5206\u5185\u5bb9\uff1a\n\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a124\n\u5207\u5206\u5185\u5bb9\uff1a\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c3\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a131\n\u5207\u5206\u5185\u5bb9\uff1a\n2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\u53ca\u65f6\u53d1\u73b0\u95ee\u9898\u5e76\u4f18\u5316\u670d\u52a1\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c4\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a125\n\u5207\u5206\u5185\u5bb9\uff1a\n3. \u6587\u672c\u751f\u6210\u4e0e\u521b\u4f5c\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u751f\u6210\u7b26\u5408\u8bed\u6cd5\u89c4\u5219\u7684\u6587\u7ae0\u3001\u65b0\u95fb\u3001\u5c0f\u8bf4\u7b49\u6587\u672c\u5185\u5bb9\u3002\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u6587\u672c\u6570\u636e\uff0c\u6a21\u578b\u53ef\u4ee5\u751f\u6210\u5177\u6709\u521b\u9020\u6027\u548c\u76f8\u5173\u6027\u7684\u5185\u5bb9\uff0c\u5e7f\u6cdb\u5e94\u7528\u4e8e\u65b0\u95fb\u62a5\u9053\u3001\u5e7f\u544a\u8425\u9500\u7b49\u9886\u57df\u3002\u6b64\u5916\uff0c\u5b83\u8fd8\u80fd\u6839\u636e\u7ed9\u5b9a\u4e3b\u9898\u6216\u5173\u952e\u8bcd\u751f\u6210\u6587\u7ae0\uff0c\u4e3a\u521b\u4f5c\u8005\u63d0\u4f9b\u7075\u611f\u548c\u8f85\u52a9\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c5\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a106\n...\n\u5207\u5206\u5185\u5bb9\uff1a\n\u7efc\u4e0a\u6240\u8ff0\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u5728\u591a\u4e2a\u9886\u57df\u90fd\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u524d\u666f\u548c\u5de8\u5927\u7684\u4ef7\u503c\u6f5c\u529b\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\u548c\u5b8c\u5584\uff0c\u6211\u4eec\u6709\u7406\u7531\u76f8\u4fe1\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u5728\u672a\u6765\u7684\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d1\u6325\u66f4\u52a0\u91cd\u8981\u7684\u4f5c\u7528\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n<\/code><\/pre>\n<pre><code class=\"language-bash\"># \u5982\u679c\u6539\u4e3achunk_size=512, chunk_overlap=32\n\u5207\u5206\u4e2a\u6570\uff1a3\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a436\n\u5207\u5206\u5185\u5bb9\uff1a\n\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\n\n2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\u53ca\u65f6\u53d1\u73b0\u95ee\u9898\u5e76\u4f18\u5316\u670d\u52a1\u3002\n\n3. \u6587\u672c\u751f\u6210\u4e0e\u521b\u4f5c\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u751f\u6210\u7b26\u5408\u8bed\u6cd5\u89c4\u5219\u7684\u6587\u7ae0\u3001\u65b0\u95fb\u3001\u5c0f\u8bf4\u7b49\u6587\u672c\u5185\u5bb9\u3002\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u6587\u672c\u6570\u636e\uff0c\u6a21\u578b\u53ef\u4ee5\u751f\u6210\u5177\u6709\u521b\u9020\u6027\u548c\u76f8\u5173\u6027\u7684\u5185\u5bb9\uff0c\u5e7f\u6cdb\u5e94\u7528\u4e8e\u65b0\u95fb\u62a5\u9053\u3001\u5e7f\u544a\u8425\u9500\u7b49\u9886\u57df\u3002\u6b64\u5916\uff0c\u5b83\u8fd8\u80fd\u6839\u636e\u7ed9\u5b9a\u4e3b\u9898\u6216\u5173\u952e\u8bcd\u751f\u6210\u6587\u7ae0\uff0c\u4e3a\u521b\u4f5c\u8005\u63d0\u4f9b\u7075\u611f\u548c\u8f85\u52a9\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a443\n\u5207\u5206\u5185\u5bb9\uff1a\n4. \u60c5\u611f\u5206\u6790\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u5206\u6790\u6587\u672c\u4e2d\u7684\u60c5\u611f\u503e\u5411\u548c\u60c5\u611f\u8868\u8fbe\uff0c\u5e2e\u52a9\u4f01\u4e1a\u4e86\u89e3\u5ba2\u6237\u53cd\u9988\u548c\u60c5\u611f\u72b6\u6001\uff0c\u4ece\u800c\u5236\u5b9a\u66f4\u7cbe\u51c6\u7684\u8425\u9500\u7b56\u7565\u6216\u4f18\u5316\u5ba2\u6237\u670d\u52a1\u3002\u8fd9\u79cd\u6280\u672f\u8fd8\u53ef\u7528\u4e8e\u793e\u4ea4\u5a92\u4f53\u76d1\u63a7\uff0c\u5b9e\u65f6\u5206\u6790\u516c\u4f17\u5bf9\u67d0\u4e00\u4e3b\u9898\u6216\u4e8b\u4ef6\u7684\u60c5\u7eea\u548c\u53cd\u5e94\u3002\n\n5. \u81ea\u52a8\u95ee\u7b54\u7cfb\u7edf\n\u63cf\u8ff0\uff1a\u901a\u8fc7\u5b66\u4e60\u5927\u91cf\u95ee\u9898\u548c\u7b54\u6848\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u81ea\u52a8\u751f\u6210\u7b26\u5408\u8bed\u6cd5\u89c4\u5219\u7684\u95ee\u9898\u548c\u7b54\u6848\u3002\u8fd9\u79cd\u81ea\u52a8\u95ee\u7b54\u7cfb\u7edf\u53ef\u5e94\u7528\u4e8e\u667a\u80fd\u52a9\u624b\u3001\u641c\u7d22\u5f15\u64ce\u7b49\u9886\u57df\uff0c\u4e3a\u7528\u6237\u63d0\u4f9b\u9ad8\u6548\u3001\u51c6\u786e\u7684\u4fe1\u606f\u670d\u52a1\u3002\u7ed3\u5408\u77e5\u8bc6\u56fe\u8c31\u6280\u672f\uff0c\u95ee\u7b54\u7cfb\u7edf\u7684\u77e5\u8bc6\u68c0\u7d22\u548c\u63a8\u7406\u80fd\u529b\u5f97\u5230\u8fdb\u4e00\u6b65\u589e\u5f3a\u3002\n\n6. \u81ea\u52a8\u6458\u8981\u4e0e\u603b\u7ed3\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u80fd\u591f\u81ea\u52a8\u5bf9\u6587\u672c\u8fdb\u884c\u6458\u8981\u548c\u603b\u7ed3\uff0c\u63d0\u53d6\u5173\u952e\u4fe1\u606f\uff0c\u5e2e\u52a9\u7528\u6237\u5feb\u901f\u4e86\u89e3\u6587\u672c\u4e3b\u65e8\u548c\u91cd\u70b9\u3002\u8fd9\u79cd\u6280\u672f\u5728\u5b66\u672f\u8bba\u6587\u3001\u65b0\u95fb\u62a5\u9053\u7b49\u9886\u57df\u5177\u6709\u91cd\u8981\u5e94\u7528\u4ef7\u503c\uff0c\u63d0\u9ad8\u4e86\u4fe1\u606f\u83b7\u53d6\u7684\u6548\u7387\u3002\n...\n\n\u7efc\u4e0a\u6240\u8ff0\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u5728\u591a\u4e2a\u9886\u57df\u90fd\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u524d\u666f\u548c\u5de8\u5927\u7684\u4ef7\u503c\u6f5c\u529b\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\u548c\u5b8c\u5584\uff0c\u6211\u4eec\u6709\u7406\u7531\u76f8\u4fe1\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u5728\u672a\u6765\u7684\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d1\u6325\u66f4\u52a0\u91cd\u8981\u7684\u4f5c\u7528\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n<\/code><\/pre>\n<h3><span class=\"ez-toc-section\" id=\"%E9%80%92%E5%BD%92%E6%96%B9%E5%BC%8F%E5%88%87%E5%88%86\"><\/span>\u9012\u5f52\u65b9\u5f0f\u5207\u5206<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"RecursiveCharacterTextSplitter%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>RecursiveCharacterTextSplitter\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_text_splitters.character.RecursiveCharacterTextSplitter(\n    separators: Optional[List[str]] = None,\n    keep_separator: bool = True,\n    is_separator_regex: bool = False,\n    **kwargs: Any\n)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>separators\uff1a\u7528\u4e8e\u6307\u5b9a\u5207\u5206\u6587\u672c\u65f6\u4f7f\u7528\u7684\u5206\u9694\u7b26\u5217\u8868\u3002\u53ef\u4ee5\u5305\u62ec\u6362\u884c\u7b26\u3001\u53e5\u53f7\u3001\u9017\u53f7\u7b49\u3002<\/li>\n<li>keep_separator\uff1a\u6307\u5b9a\u5728\u5207\u5206\u540e\u662f\u5426\u4fdd\u7559\u5206\u9694\u7b26\u3002\u5982\u679c\u8bbe\u7f6e\u4e3a True\uff0c\u5207\u5206\u540e\u7684\u6587\u672c\u7247\u6bb5\u5c06\u5305\u542b\u5206\u9694\u7b26\u3002<\/li>\n<li>is_separator_regex\uff1a\u6307\u5b9a separators \u662f\u5426\u4e3a\u6b63\u5219\u8868\u8fbe\u5f0f\u3002\u5982\u679c\u8bbe\u7f6e\u4e3a True\uff0c\u5219 separators \u5c06\u88ab\u89c6\u4e3a\u6b63\u5219\u8868\u8fbe\u5f0f\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-11\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import RecursiveCharacterTextSplitter\n\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\n\ndocs = text_loader.load()\n\nseparators=[&quot;\\n&quot;, &quot;.&quot;, &quot;\u3002&quot;, &quot;!&quot;, &quot;\uff01&quot;, &quot;?&quot;, &quot;\uff1f&quot;, &quot;\uff1b&quot;, &quot;;&quot;, &quot;\u2026\u2026&quot;, &quot;\u2026&quot;, &quot;\u3001&quot;, &quot;\uff0c&quot;, &quot;,&quot;, &quot; &quot;]\n\n# \u521b\u5efa RecursiveCharacterTextSplitter \u5b9e\u4f8b\nsplitter = RecursiveCharacterTextSplitter(\n    separators=separators,\n    keep_separator=True,\n    is_separator_regex=False,\n    chunk_size=128,\n    chunk_overlap=32\n)\n\nresults = splitter.split_documents(documents=docs)\n\n# \u6253\u5370\u5207\u5206\u7ed3\u679c\nprint(f&#039;\u5207\u5206\u4e2a\u6570\uff1a{len(results)}&#039;)\n\nfor idx, chunk in enumerate(results):\n    print(f&#039;\u7b2c{idx+1}\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a{len(chunk.page_content)}&#039;)\n    print(f&#039;\u5207\u5206\u5185\u5bb9\uff1a\\n{chunk.page_content}&#039;)\n    print(&quot;-*-&quot; *30 + &quot;\\n&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">\u5207\u5206\u4e2a\u6570\uff1a13\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a59\n\u5207\u5206\u5185\u5bb9\uff1a\n\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n\n1. \u673a\u5668\u7ffb\u8bd1\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a124\n\u5207\u5206\u5185\u5bb9\uff1a\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c3\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a13\n\u5207\u5206\u5185\u5bb9\uff1a\n2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c4\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a117\n\u5207\u5206\u5185\u5bb9\uff1a\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\u53ca\u65f6\u53d1\u73b0\u95ee\u9898\u5e76\u4f18\u5316\u670d\u52a1\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c5\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a125\n...\n\u5207\u5206\u5185\u5bb9\uff1a\n\u7efc\u4e0a\u6240\u8ff0\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u5728\u591a\u4e2a\u9886\u57df\u90fd\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u524d\u666f\u548c\u5de8\u5927\u7684\u4ef7\u503c\u6f5c\u529b\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\u548c\u5b8c\u5584\uff0c\u6211\u4eec\u6709\u7406\u7531\u76f8\u4fe1\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u5728\u672a\u6765\u7684\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d1\u6325\u66f4\u52a0\u91cd\u8981\u7684\u4f5c\u7528\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-<\/code><\/pre>\n<p>\u5bf9\u6bd4CharacterTextSplitter\u7684\u6267\u884c\u7ed3\u679c\uff0cRecursiveCharacterTextSplitter\u5728\u5207\u5206\u957f\u6587\u672c\u65f6\uff0c\u53ef\u4ee5\u5207\u5206\u51fa\u66f4\u5c0f\u7684\u6587\u672c\u7247\u6bb5\uff0c\u5e76\u4e14\u53ef\u4ee5\u4fdd\u7559\u5206\u9694\u7b26\u3002<\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E6%8C%89%E7%85%A7Token%E5%88%87%E5%88%86\"><\/span>\u6309\u7167Token\u5207\u5206<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"TokenTextSplitter%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E\"><\/span>TokenTextSplitter\u51fd\u6570\u8bf4\u660e<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">class langchain_text_splitters.base.TokenTextSplitter(\n    encoding_name: str = &#039;gpt2&#039;,\n    model_name: Optional[str] = None,\n    allowed_special: Union[Literal[&#039;all&#039;], AbstractSet[str]] = {},\n    disallowed_special: Union[Literal[&#039;all&#039;], Collection[str]] = &#039;all&#039;,\n    **kwargs: Any\n)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>model_name(\u53ef\u9009)\uff1a\u53ef\u9009\u53c2\u6570\uff0c\u7528\u4e8e\u6307\u5b9a\u7279\u5b9a\u7684\u8bed\u8a00\u6a21\u578b\u540d\u79f0\u3002<\/li>\n<li>allowed_special\uff1a\u6307\u5b9a\u88ab\u5141\u8bb8\u7684\u7279\u6b8a\u4ee4\u724c\u96c6\u5408\u3002<\/li>\n<li>disallowed_special\uff1a\u6307\u5b9a\u4e0d\u5141\u8bb8\u7684\u7279\u6b8a\u4ee4\u724c\u96c6\u5408\u3002<\/li>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u4f9d\u8d56\uff1a<code>pip install tiktoken<\/code>\u3002<\/li>\n<li>encoding_name\uff1a\u6307\u5b9a\u7528\u4e8e\u4ee4\u724c\u5316\u6587\u672c\u7684\u7f16\u7801\u540d\u79f0\u3002\u8be5\u9009\u9879\u901a\u8fc7\u4ee5\u4e0b\u4ee3\u7801\u53ef\u4ee5\u5f97\u5230\u5e38\u89c1\u9009\u9879\uff1a<\/li>\n<\/ul>\n<pre><code class=\"language-python\">import tiktoken\ntiktoken.list_encoding_names()\n\n# \u8fd0\u884c\u7ed3\u679c\uff1a\n# [&#039;gpt2&#039;, &#039;r50k_base&#039;, &#039;p50k_base&#039;, &#039;p50k_edit&#039;, &#039;cl100k_base&#039;, &#039;o200k_base&#039;]<\/code><\/pre>\n<ul>\n<li><code>gpt2<\/code>:\u8fd9\u662f OpenAI \u7684 <code>GPT-2<\/code> \u6a21\u578b\u4f7f\u7528\u7684\u7f16\u7801\uff0c\u57fa\u4e8e <code>Byte Pair Encoding (BPE)<\/code>\u3002\u5b83\u9002\u7528\u4e8e\u5927\u591a\u6570\u57fa\u4e8e GPT-2 \u7684\u4efb\u52a1\uff0c\u80fd\u591f\u6709\u6548\u5904\u7406<strong>\u82f1\u8bed\u6587\u672c<\/strong>\u3002<\/li>\n<li><code>r50k_base<\/code>:\u8fd9\u662f\u9488\u5bf9 <code>OpenAI<\/code> \u7684\u6a21\u578b\uff08\u5982 <code>GPT-3<\/code>\uff09\u7684\u4e00\u79cd\u7f16\u7801\uff0c\u4f7f\u7528\u4e86 <code>50,000<\/code> \u4e2a\u6700\u5e38\u89c1\u7684\u5b50\u8bcd\u5355\u4f4d\u3002\u5b83\u662f\u4e00\u79cd\u66f4\u4e3a\u7cbe\u7b80\u7684\u7f16\u7801\u65b9\u5f0f\uff0c\u9002\u7528\u4e8e\u5bf9\u6587\u672c\u8fdb\u884c\u66f4\u7ec6\u7c92\u5ea6\u7684\u5207\u5206\u3002<\/li>\n<li><code>p50k_base<\/code>:\u57fa\u4e8e <code>BPE<\/code> \u7684\u7f16\u7801\uff0c\u4f7f\u7528 <code>50,000<\/code> \u4e2a\u5b50\u8bcd\u5355\u4f4d\u3002\u8be5\u7f16\u7801\u4e3b\u8981\u7528\u4e8e\u5904\u7406\u66f4\u590d\u6742\u7684\u6587\u672c\u751f\u6210\u548c\u7406\u89e3\u4efb\u52a1\uff0c\u9002\u5408\u4e0e <code>GPT-3<\/code> \u53ca\u76f8\u5173\u6a21\u578b\u7684\u914d\u5408\u4f7f\u7528\u3002<\/li>\n<li><code>p50k_edit<\/code>:\u8fd9\u662f <code>p50k_base<\/code> \u7684\u4e00\u79cd\u53d8\u4f53\uff0c\u4e13\u95e8\u4f18\u5316\u7528\u4e8e\u6587\u672c\u7f16\u8f91\u4efb\u52a1\u3002\u8fd9\u79cd\u7f16\u7801\u5173\u6ce8\u4e8e\u6587\u672c\u7684\u4fee\u6539\u548c\u8c03\u6574\uff0c\u9002\u5408\u9700\u8981\u5bf9\u6587\u672c\u8fdb\u884c\u7ec6\u5fae\u7f16\u8f91\u7684\u5e94\u7528\u573a\u666f\u3002<\/li>\n<li><code>cl100k_base<\/code>:\u8fd9\u662f\u9488\u5bf9\u66f4\u5927\u89c4\u6a21\u6a21\u578b\uff08\u5982 <code>GPT-4<\/code>\uff09\u7684\u4e00\u79cd\u7f16\u7801\uff0c\u4f7f\u7528\u4e86 <code>100,000<\/code> \u4e2a\u5b50\u8bcd\u5355\u4f4d\u3002\u5b83\u80fd\u591f\u5904\u7406\u66f4\u590d\u6742\u548c\u591a\u6837\u5316\u7684\u6587\u672c\uff0c\u9002\u5408\u9ad8\u6027\u80fd\u7684\u6587\u672c\u751f\u6210\u548c\u7406\u89e3\u4efb\u52a1\u3002<\/li>\n<li><code>o200k_base<\/code>:\u8fd9\u662f\u4e00\u4e2a\u66f4\u9ad8\u7ea7\u7684\u7f16\u7801\uff0c\u4f7f\u7528\u4e86 <code>200,000<\/code> \u4e2a\u5b50\u8bcd\u5355\u4f4d\uff0c\u65e8\u5728\u5904\u7406\u6781\u4e3a\u590d\u6742\u7684\u6587\u672c\u4efb\u52a1\u3002\u5b83\u9002\u5408\u9700\u8981\u5904\u7406\u5927\u91cf\u4fe1\u606f\u6216\u591a\u6837\u5316\u6587\u672c\u7684\u5e94\u7528\uff0c\u80fd\u591f\u63d0\u4f9b\u66f4\u9ad8\u7684\u7075\u6d3b\u6027\u548c\u7cbe\u5ea6\u3002<\/li>\n<\/ul>\n<blockquote>\n<p>\u4ee5\u4e0a\u9009\u9879\u5185\u5bb9\u7531ChatGPT\u8f85\u52a9\u751f\u6210\uff0c\u5982\u6709\u9519\u8bef\uff0c\u8bf7\u53ca\u65f6\u53cd\u9988\u3002<\/p>\n<\/blockquote>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-12\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">import tiktoken\nfrom langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import TokenTextSplitter\n\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\n\ndocs = text_loader.load()\n\nsplitter = TokenTextSplitter(encoding_name=&quot;o200k_base&quot;, \n                     chunk_size=128, \n                     chunk_overlap=64)\n\nresults = splitter.split_documents(documents=docs)\n\n# \u6253\u5370\u5207\u5206\u7ed3\u679c\nprint(f&#039;\u5207\u5206\u4e2a\u6570\uff1a{len(results)}&#039;)\n\nfor idx, chunk in enumerate(results):\n    print(f&#039;\u7b2c{idx+1}\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a{len(chunk.page_content)}&#039;)\n    print(f&#039;\u5207\u5206\u5185\u5bb9\uff1a\\n{chunk.page_content}&#039;)\n    print(&quot;-*-&quot; *30 + &quot;\\n&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">\u5207\u5206\u4e2a\u6570\uff1a13\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a185\n\u5207\u5206\u5185\u5bb9\uff1a\n\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\n\n2. \u667a\u80fd\u5ba2\u670d\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a198\n\u5207\u5206\u5185\u5bb9\uff1a\n\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002\n\n2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c3\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a215\n\u5207\u5206\u5185\u5bb9\uff1a\n\u4e0e\u804a\u5929\u673a\u5668\u4eba\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u88ab\u7528\u4e8e\u5f00\u53d1\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u548c\u804a\u5929\u673a\u5668\u4eba\uff0c\u80fd\u591f\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\u5e76\u63d0\u4f9b\u76f8\u5e94\u7684\u89e3\u51b3\u65b9\u6848\u6216\u8f6c\u8fbe\u7ed9\u76f8\u5173\u90e8\u95e8\u3002\u8fd9\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u5ba2\u670d\u6548\u7387\uff0c\u8fd8\u63d0\u5347\u4e86\u7528\u6237\u4f53\u9a8c\u3002\u4f8b\u5982\uff0c\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\u6280\u672f\uff0c\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u53ef\u4ee5\u5206\u6790\u7528\u6237\u60c5\u611f\u72b6\u6001\uff0c\u53ca\u65f6\u53d1\u73b0\u95ee\u9898\u5e76\u4f18\u5316\u670d\u52a1\u3002\n\n3. \u6587\u672c\u751f\u6210\u4e0e\u521b\u4f5c\n...\n\n\u7efc\u4e0a\u6240\u8ff0\uff0c\u5927\u8bed\u8a00\u6a21\u578b\u5728\u591a\u4e2a\u9886\u57df\u90fd\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u524d\u666f\u548c\u5de8\u5927\u7684\u4ef7\u503c\u6f5c\u529b\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\u548c\u5b8c\u5584\uff0c\u6211\u4eec\u6709\u7406\u7531\u76f8\u4fe1\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u5728\u672a\u6765\u7684\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d1\u6325\u66f4\u52a0\u91cd\u8981\u7684\u4f5c\u7528\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n<\/code><\/pre>\n<p>\u5982\u679c\u5c06encoding_name\u8bbe\u7f6e\u4e3a&#8217;gpt2&#8217;\uff0c\u5219\u5207\u5206\u7ed3\u679c\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-bash\">\u5207\u5206\u4e2a\u6570\uff1a43\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a65\n\u5207\u5206\u5185\u5bb9\uff1a\n\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\ufffd\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a62\n\u5207\u5206\u5185\u5bb9\uff1a\n\ufffd\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a\n\n1. \u673a\u5668\u7ffb\u8bd1\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c3\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a58\n\u5207\u5206\u5185\u5bb9\uff1a\n\ufffd\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\ufffd\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n<\/code><\/pre>\n<p>\u901a\u8fc7\u5bf9\u6bd4\uff0c\u53ef\u4ee5\u770b\u5230o200k_base\u7684\u6548\u679c\u8981\u4f18\u4e8egpt2\u3002<\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E6%8C%89%E7%85%A7markdown%E6%A0%87%E9%A2%98%E5%88%87%E5%88%86\"><\/span>\u6309\u7167markdown\u6807\u9898\u5207\u5206<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"%E7%A4%BA%E4%BE%8B-13\"><\/span>\u793a\u4f8b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import MarkdownTextSplitter\n\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.md&quot;, encoding=&quot;utf8&quot;)\n\ndocs = text_loader.load()\n\n# \u521b\u5efa MarkdownTextSplitter \u5b9e\u4f8b\nsplitter = MarkdownTextSplitter(chunk_size=128, chunk_overlap=64)\n\nresults = splitter.split_documents(documents=docs)\n\n# \u6253\u5370\u5207\u5206\u7ed3\u679c\nprint(f&#039;\u5207\u5206\u4e2a\u6570\uff1a{len(results)}&#039;)\n\nfor idx, chunk in enumerate(results):\n    print(f&#039;\u7b2c{idx+1}\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a{len(chunk.page_content)}&#039;)\n    print(f&#039;\u5207\u5206\u5185\u5bb9\uff1a\\n{chunk.page_content}&#039;)\n    print(&quot;-*-&quot; *30 + &quot;\\n&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-markdown\">\u5207\u5206\u4e2a\u6570\uff1a196\n\u7b2c1\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a126\n\u5207\u5206\u5185\u5bb9\uff1a\n## \u524d\u8a00\n\u672c\u7ae0\u6211\u4eec\u5c06\u901a\u8fc7 LLaMA-Factory \u5177\u4f53\u5b9e\u8df5\u5927\u6a21\u578b\u8bad\u7ec3\u7684\u4e09\u4e2a\u9636\u6bb5\uff0c\u5305\u62ec\uff1a\u9884\u8bad\u7ec3\u3001\u76d1\u7763\u5fae\u8c03\u548c\u504f\u597d\u7ea0\u6b63\u3002\n\n## \u5927\u6a21\u578b\u8bad\u7ec3\u56de\u987e\n![](\u5927\u6a21\u578b\u8bad\u7ec3\u7684\u4e09\u4e2a\u9636\u6bb5.png)\n\n## \u8bad\u7ec3\u76ee\u6807\n\u901a\u8fc7\u5b9e\u8df5\u5927\u6a21\u578b\u7684\u4e09\u4e2a\u8bad\u7ec3\u9636\u6bb5\uff0c\u8bad\u7ec3\u4e00\u4e2a\u533b\u7597\u5927\u6a21\u578b\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c2\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a123\n\u5207\u5206\u5185\u5bb9\uff1a\n## \u8bad\u7ec3\u8fc7\u7a0b\u5b9e\u65bd\n### \u51c6\u5907\u8bad\u7ec3\u6846\u67b6\n`LLaMA Factory`\u662f\u4e00\u6b3e\u5f00\u6e90\u4f4e\u4ee3\u7801\u5927\u6a21\u578b\u5fae\u8c03\u6846\u67b6\uff0c\u96c6\u6210\u4e86\u4e1a\u754c\u6700\u5e7f\u6cdb\u4f7f\u7528\u7684\u5fae\u8c03\u6280\u672f\uff0c\u652f\u6301\u901a\u8fc7Web UI\u754c\u9762\u96f6\u4ee3\u7801\u5fae\u8c03\u5927\u6a21\u578b\uff0c\u76ee\u524d\u5df2\u7ecf\u6210\u4e3a\u5f00\u6e90\u793e\u533a\u5185\u6700\u53d7\u6b22\u8fce\u7684\u5fae\u8c03\u6846\u67b6\uff0cGitHub\u661f\u6807\u8d85\u8fc72\u4e07\u3002\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-\n\n\u7b2c3\u6bb5\uff0c\u957f\u5ea6\u4e3a\uff1a118\n\u5207\u5206\u5185\u5bb9\uff1a\n#### \u8fd0\u884c\u73af\u5883\u8981\u6c42\n- \u786c\u4ef6\uff1a\n  - GPU\uff1a\u63a8\u8350\u4f7f\u752824GB\u663e\u5b58\u7684\u663e\u5361\u6216\u8005\u66f4\u9ad8\u914d\u7f6e\n...\n## \u53c2\u8003\u8d44\u6599\n[Github\uff1aLLaMA-Factory\u7684README\u6587\u4ef6](https:\/\/github.com\/hiyouga\/LLaMA-Factory\/blob\/main\/data\/README.md)\n-*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*-<\/code><\/pre>\n<p>\u5bf9\u4e8eMarkdown\u6587\u672c\uff0c\u4f7f\u7528MarkdownTextSplitter\u8fdb\u884c\u5207\u5206\uff0c\u5207\u5206\u6548\u679c\u6bd4\u8f83\u597d\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"%E5%90%91%E9%87%8F%E5%8C%96EMBED\"><\/span>\u5411\u91cf\u5316(EMBED)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>\u8003\u8651\u5230RAG\u7cfb\u7edf\u4e00\u822c\u90fd\u6709\u79c1\u6709\u5316\u90e8\u7f72\u7684\u9700\u6c42(\u56e0\u4e3a\u6570\u636e\u6d89\u5bc6)\uff0c\u6240\u4ee5\u5bf9\u5e94\u7684\u5411\u91cf\u5316\u7684\u65b9\u5f0f\u6709\u4e24\u79cd\uff1a\u672c\u5730\u5316 \u548c \u5728\u7ebf\u5316\u3002<\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E6%9C%AC%E5%9C%B0%E5%8C%96\"><\/span>\u672c\u5730\u5316<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>\u5728\u5f00\u6e90\u9879\u76ee<a href=\"https:\/\/github.com\/netease-youdao\/QAnything\/blob\/master\/qanything_kernel\/connector\/embedding\/embedding_for_local.py\">QAnything<\/a>\u4e2d\uff0c\u5176\u63d0\u4f9b\u4e86\u4e00\u79cd\u4f7f\u7528\u672c\u5730\u6a21\u578b\u7684\u65b9\u6cd5\uff0c\u5176\u5927\u81f4\u6b65\u9aa4\u662f\uff1a<\/p>\n<ul>\n<li>\u5728\u672c\u5730\u4fdd\u5b58\u4e86\u4e00\u4e2a\u5411\u91cf\u5316\u6a21\u578b\u3002<\/li>\n<li>\u5728\u8c03\u7528\u5411\u91cf\u5316\u6a21\u578b\u65f6\uff0c\u4f7f\u7528\u672c\u5730\u7684\u6a21\u578b.<\/li>\n<\/ul>\n<pre><code class=\"language-python\">embedding_client = EmbeddingClient(\n    server_url=LOCAL_EMBED_SERVICE_URL,\n    model_name=LOCAL_EMBED_MODEL_NAME,\n    model_version=&#039;1&#039;,\n    resp_wait_s=120,\n    tokenizer_path=&#039;qanything_kernel\/connector\/embedding\/embedding_model_0630&#039;)\n<\/code><\/pre>\n<p>\u7531\u4e8e\u793a\u4f8b\u4ee3\u7801\u7684\u96c6\u6210\u5ea6\u8f83\u9ad8\uff0c\u5265\u79bb\u6838\u5fc3\u4ee3\u7801\u6267\u884c\u6210\u672c\u8f83\u5927\uff0c\u6240\u4ee5\u6b64\u5904\u5e76\u672a\u5b9e\u8df5\uff0c\u53ea\u662f\u5f15\u7528\u76f8\u5173\u7684\u4fe1\u606f\u3002<\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E5%9C%A8%E7%BA%BF%E5%8C%96\"><\/span>\u5728\u7ebf\u5316<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"%E4%BD%BF%E7%94%A8HuggingFaceEmbeddings\"><\/span>\u4f7f\u7528HuggingFaceEmbeddings<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>LangChain\u793e\u533a\u63d0\u4f9b\u4e86HuggingFaceEmbeddings\uff0c\u8be5\u7c7b\u53ef\u4ee5\u52a0\u8f7dHuggingFace\u7684\u9884\u8bad\u7ec3\u6a21\u578b\uff0c\u5e76\u4f7f\u7528\u8be5\u6a21\u578b\u8fdb\u884c\u5411\u91cf\u5316\u3002<\/p>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import RecursiveCharacterTextSplitter\nfrom langchain_chroma import Chroma\nfrom langchain.embeddings import HuggingFaceEmbeddings\n\n# \u52a0\u8f7d\u6587\u672c\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\ndocs = text_loader.load()\n\n# \u5207\u5206\u6587\u672c\nspliter = RecursiveCharacterTextSplitter(chunk_size=128, chunk_overlap=64)\ndocs = spliter.split_documents(documents=docs)\n\n# \u521d\u59cb\u5316 HuggingFaceEmbeddings\nembedding_function = HuggingFaceEmbeddings(model_name=&quot;bert-base-chinese&quot;)\n\n# \u521b\u5efa\u5411\u91cf\u6570\u636e\u5e93\nstore = Chroma(embedding_function=embedding_function, persist_directory=&quot;chroma_data&quot;)\n\n# \u5c06\u6587\u6863\u6dfb\u52a0\u5230\u6570\u636e\u5e93\u4e2d\nstore.add_documents(documents=docs)\n\n# \u8fdb\u884c\u76f8\u4f3c\u6027\u641c\u7d22\nstore.similarity_search(query=&quot;\u5c0f\u7c73\u4f30\u503c\u591a\u5c11\uff1f&quot;, k=2)\n<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u4f7f\u7528\u524d\u9700\u8981\u5b89\u88c5\u4f9d\u8d56<code>pip install sentence-transformers<\/code><\/li>\n<\/ul>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">[Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;1. \u673a\u5668\u7ffb\u8bd1\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002&#039;),\n Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;1. \u673a\u5668\u7ffb\u8bd1\\n\u63cf\u8ff0\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u901a\u8fc7\u8bad\u7ec3\u53ef\u4ee5\u5b66\u4e60\u4e0d\u540c\u8bed\u8a00\u4e4b\u95f4\u7684\u8bed\u6cd5\u548c\u8bed\u4e49\u89c4\u5219\uff0c\u5b9e\u73b0\u81ea\u52a8\u7ffb\u8bd1\u3002\u8fd9\u79cd\u6280\u672f\u5df2\u5e7f\u6cdb\u5e94\u7528\u4e8e\u8de8\u56fd\u4f01\u4e1a\u6c9f\u901a\u3001\u56fd\u9645\u5408\u4f5c\u7b49\u9886\u57df\uff0c\u5982\u8c37\u6b4c\u7ffb\u8bd1\u7b49\u4ea7\u54c1\u3002\u5c3d\u7ba1\u5728\u5904\u7406\u957f\u53e5\u5b50\u548c\u6b67\u4e49\u6d88\u89e3\u7b49\u65b9\u9762\u4ecd\u9762\u4e34\u6311\u6218\uff0c\u4f46\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u51c6\u786e\u6027\u548c\u6d41\u7545\u5ea6\u4e0d\u65ad\u63d0\u5347\u3002&#039;)]<\/code><\/pre>\n<h4><span class=\"ez-toc-section\" id=\"%E4%BD%BF%E7%94%A8QianfanEmbeddings\"><\/span>\u4f7f\u7528QianfanEmbeddings<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u7531\u4e8eQwen\u7684API\u6709\u6837\u672c\u4e3a6\u4e2a\u7684\u9650\u5236\uff0c\u6240\u4ee5\u6b64\u5904\u8bd5\u7528\u767e\u5ea6\u5343\u5e06\u5927\u6a21\u578b\u7684\u5411\u91cf\u5316API\u3002<\/p>\n<pre><code class=\"language-python\"># util.py\ndef get_ernie_models():\n    &quot;&quot;&quot;\n    \u52a0\u8f7d\u6587\u5fc3\u7cfb\u5217\u5927\u6a21\u578b\n    &quot;&quot;&quot;\n    # LLM \u5927\u8bed\u8a00\u6a21\u578b\uff08\u5355\u8f6e\u5bf9\u8bdd\u7248\uff09\n    from langchain_community.llms import QianfanLLMEndpoint\n\n    # Chat \u804a\u5929\u7248\u5927\u6a21\u578b\uff08\u652f\u6301\u591a\u8f6e\u5bf9\u8bdd\uff09\n    from langchain_community.chat_models import QianfanChatEndpoint\n\n    # Embeddings \u5d4c\u5165\u6a21\u578b\n    from langchain_community.embeddings import QianfanEmbeddingsEndpoint\n\n    llm = QianfanLLMEndpoint(model=&quot;ERNIE-Bot-turbo&quot;, temperature=0.1, top_p=0.2)\n    chat = QianfanChatEndpoint(model=&quot;ERNIE-Lite-8K&quot;, top_p=0.2, temperature=0.1)\n\n    embed = QianfanEmbeddingsEndpoint(model=&quot;bge-large-zh&quot;)\n\n    return llm, chat, embed<\/code><\/pre>\n<pre><code class=\"language-python\">from langchain_community.document_loaders import TextLoader\nfrom langchain_text_splitters import RecursiveCharacterTextSplitter\nfrom langchain_chroma import Chroma\nfrom utils import get_ernie_models\n\n# \u8fde\u63a5\u5927\u6a21\u578b\nllm_ernie, chat_ernie, embed_ernie = get_ernie_models()\n\n# \u52a0\u8f7d\u6587\u672c\ntext_loader = TextLoader(file_path=&quot;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&quot;, encoding=&quot;utf8&quot;)\ndocs = text_loader.load()\n\n# \u5207\u5206\u6587\u672c\nspliter = RecursiveCharacterTextSplitter(chunk_size=128, chunk_overlap=64)\ndocs = spliter.split_documents(documents=docs)\n\n# \u4f7f\u7528 ernie \u6a21\u578b\u521b\u5efa\u5411\u91cf\u6570\u636e\u5e93\nstore = Chroma.from_documents(documents=docs, \n                              embedding=embed_ernie, \n                              persist_directory=&quot;chroma_data2&quot;)\n\nstore.similarity_search(query=&quot;\u5c0f\u7c73\u4f30\u503c\u591a\u5c11\uff1f&quot;, k=2)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">[Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba&#039;),\n Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a&#039;)]<\/code><\/pre>\n<p>\u901a\u8fc7\u5bf9\u6bd4<\/p>\n<ul>\n<li>\u4f7f\u7528QianfanEmbeddingsEndpoint\u5411\u91cf\u5316\u540e\uff0c\u5728\u76f8\u4f3c\u6027\u641c\u7d22\u65f6\u6709\u4e24\u4e2a\u5185\u5bb9\u8fd4\u56de\uff1b\u800c\u4f7f\u7528HuggingFaceEmbeddings\u5411\u91cf\u5316\u540e\uff0c\u5728\u76f8\u4f3c\u6027\u641c\u7d22\u65f6\u53ea\u6709\u7b2c\u4e00\u4e2a\u5185\u5bb9\u8fd4\u56de\u3002<\/li>\n<li>\u76f8\u4f3c\u6027\u641c\u7d22\u540e\u7684\u5185\u5bb9\uff0c\u770b\u8d77\u6765\u4e0e\u6211\u4eecQuery\u67e5\u8be2\u7684\u95ee\u9898\u4f3c\u4e4e\u4e0d\u592a\u76f8\u5173\uff0c\u53ea\u662f\u6a21\u578b\u5728\u5411\u91cf\u5c42\u9762\u8ba1\u7b97\u662f\u6bd4\u8f83\u76f8\u4f3c\u3002<\/li>\n<\/ul>\n<h3><span class=\"ez-toc-section\" id=\"%E5%8E%9F%E7%90%86%E6%8E%A2%E5%AF%BB\"><\/span>\u539f\u7406\u63a2\u5bfb<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<pre><code class=\"language-python\">import numpy as np\nfrom utils import get_ernie_models\n\nllm_ernie, chat_ernie, embed_ernie = get_ernie_models()\n\nquery = np.array(embed_ernie.embed_query(text=&quot;\u5c0f\u7c73\u4f30\u503c\u591a\u5c11\uff1f&quot;))\ndoc1 = np.array(embed_ernie.embed_query(text=&quot;2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba&quot;))\ndoc2 = np.array(embed_ernie.embed_query(text=&quot;\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a&quot;))\n\n# \u6b27\u5f0f\u8ddd\u79bb\nresult1 = ((query - doc1) ** 2).sum()\nprint(f&#039;query \u548c doc1 \u7684\u6b27\u5f0f\u8ddd\u79bb\u4e3a\uff1a{result1}&#039;)\n\nresult2 =((query - doc2) ** 2).sum()\nprint(f&#039;query \u548c doc2 \u7684\u6b27\u5f0f\u8ddd\u79bb\u4e3a\uff1a{result2}&#039;)\n\n# \u76f8\u5173\u7cfb\u6570\ncorr1  = 1 - result1 \/ 2 ** 0.5\nprint(f&#039;query \u548c doc1 \u7684\u76f8\u5173\u7cfb\u6570\u4e3a\uff1a{corr1}&#039;)\n\ncorr2 = 1 - result2 \/ 2 ** 0.5\nprint(f&#039;query \u548c doc2 \u7684\u76f8\u5173\u7cfb\u6570\u4e3a\uff1a{corr2}&#039;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\"># query \u548c doc1 \u7684\u6b27\u5f0f\u8ddd\u79bb\u4e3a\uff1a0.6204636182956037\n# query \u548c doc2 \u7684\u6b27\u5f0f\u8ddd\u79bb\u4e3a\uff1a0.8112505547683875\n# query \u548c doc1 \u7684\u76f8\u5173\u7cfb\u6570\u4e3a\uff1a0.561265968023637\n# query \u548c doc2 \u7684\u76f8\u5173\u7cfb\u6570\u4e3a\uff1a0.42635923148192456<\/code><\/pre>\n<p>\u5bf9\u6bd4\u5982\u4e0b\u4ee3\u7801\u6267\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-python\"># \u76f8\u5173\u6027\u8f6c\u6362\nstore.similarity_search_with_relevance_scores(query=&quot;\u5c0f\u7c73\u4f30\u503c\u591a\u5c11\uff1f&quot;, k=2)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">[(Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;2. \u667a\u80fd\u5ba2\u670d\u4e0e\u804a\u5929\u673a\u5668\u4eba&#039;),\n  0.5612658250623314),\n (Document(metadata={&#039;source&#039;: &#039;testfiles\/\u6d4b\u8bd5\u6587\u4ef6.txt&#039;}, page_content=&#039;\u5927\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u91cd\u8981\u6280\u672f\u4e4b\u4e00\uff0c\u5177\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\u3002\u4ee5\u4e0b\u662f\u5341\u4e2a\u65b9\u9762\u7684\u5e94\u7528\u573a\u666f\u53ca\u5176\u8be6\u7ec6\u63cf\u8ff0\uff1a&#039;),\n  0.4263594055624078)]<\/code><\/pre>\n<p>\u7531\u4e0a\u53ef\u4ee5\u770b\u5230\uff1a<\/p>\n<ul>\n<li>\u76f8\u4f3c\u5ea6\u8ba1\u7b97\uff1a\u901a\u8fc7\u8ba1\u7b97\u4e24\u4e2a\u5411\u91cf\u7684\u6b27\u5f0f\u8ddd\u79bb\uff0c\u53ef\u4ee5\u5f97\u5230\u4e24\u4e2a\u5411\u91cf\u4e4b\u95f4\u7684\u76f8\u4f3c\u5ea6\uff0c\u8ba1\u7b97\u7ed3\u679c\u8d8a\u5c0f\u4ee3\u8868\u8d8a\u63a5\u8fd1\u3002<\/li>\n<li>\u76f8\u4f3c\u5ea6\u8f6c\u6362\uff1a\u901a\u8fc7\u8ba1\u7b97\u4e24\u4e2a\u5411\u91cf\u7684\u76f8\u5173\u7cfb\u6570\uff0c\u53ef\u4ee5\u5f97\u5230\u4e24\u4e2a\u5411\u91cf\u4e4b\u95f4\u7684\u76f8\u4f3c\u5ea6\uff0c\u8ba1\u7b97\u7ed3\u679c\u8d8a\u5927\u4ee3\u8868\u8d8a\u76f8\u4f3c\u3002<\/li>\n<li>\u6b27\u5f0f\u8ddd\u79bb\u8ba1\u7b97\u65b9\u6cd5\uff1a$d(\\mathbf{a}, \\mathbf{b}) = \\sqrt{\\sum_{i=1}^{n} (a_i &#8211; b_i)^2}$<\/li>\n<li>\u76f8\u5173\u7cfb\u6570\u8ba1\u7b97\u65b9\u6cd5\uff1a$r = 1 &#8211; \\frac{d(\\mathbf{query}, \\mathbf{doc})}{\\sqrt{2}}$<\/li>\n<li><code>similarity_search<\/code> \u5373\u4e3a\u76f8\u4f3c\u5ea6\u8ba1\u7b97\u3002<\/li>\n<li><code>similarity_search_with_relevance_scores<\/code> \u5373\u4e3a\u76f8\u4f3c\u5ea6\u8f6c\u6362\uff0c\u5b83\u662f <code>smimilarity_search<\/code> \u7684\u6269\u5c55\uff0c\u53ef\u4ee5\u8fd4\u56de\u76f8\u4f3c\u5ea6\u3002<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%AD%98%E5%82%A8STORE\"><\/span>\u5b58\u50a8(STORE)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>\u5982\u4e0a\u8ff0\u793a\u4f8b\uff0c\u6211\u4eec\u5bf9\u6570\u636e\u8fdb\u884c\u5411\u91cf\u5316\u4e4b\u540e\uff0c\u4e0b\u4e00\u6b65\u9700\u8981\u5b58\u50a8\u81f3\u5411\u91cf\u6570\u636e\u5e93\u3002\u76ee\u524d\u5e02\u9762\u4e0a\u7684\u5411\u91cf\u6570\u636e\u5e93\u4f17\u591a\uff0c\u4e3b\u6d41\u7684\u5411\u91cf\u6570\u636e\u5e93\u5bf9\u6bd4\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n<table>\n<thead>\n<tr>\n<th>\u5411\u91cf\u6570\u636e\u5e93<\/th>\n<th>URL<\/th>\n<th>GitHub Star<\/th>\n<th>Language<\/th>\n<th>Cloud<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>chroma<\/td>\n<td><a href=\"https:\/\/github.com\/chroma-core\/chroma\">chroma<\/a><\/td>\n<td>7.4K<\/td>\n<td>Python<\/td>\n<td>\u274c<\/td>\n<\/tr>\n<tr>\n<td>milvus<\/td>\n<td><a href=\"https:\/\/github.com\/milvus-io\/milvus\">milvus<\/a><\/td>\n<td>21.5K<\/td>\n<td>Go\/Python\/C++<\/td>\n<td>\u2705<\/td>\n<\/tr>\n<tr>\n<td>pinecone<\/td>\n<td><a href=\"https:\/\/www.pinecone.io\/\">pinecone<\/a><\/td>\n<td>\u274c<\/td>\n<td>\u274c<\/td>\n<td>\u2705<\/td>\n<\/tr>\n<tr>\n<td>qdrant<\/td>\n<td><a href=\"https:\/\/github.com\/qdrant\/qdrant\">qdrant<\/a><\/td>\n<td>11.8K<\/td>\n<td>Rust<\/td>\n<td>\u2705<\/td>\n<\/tr>\n<tr>\n<td>typesense<\/td>\n<td><a href=\"https:\/\/github.com\/typesense\/typesense\">typesense<\/a><\/td>\n<td>12.9K<\/td>\n<td>C++<\/td>\n<td>\u274c<\/td>\n<\/tr>\n<tr>\n<td>weaviate<\/td>\n<td><a href=\"https:\/\/github.com\/weaviate\/weaviate\">weaviate<\/a><\/td>\n<td>6.9K<\/td>\n<td>Go<\/td>\n<td>\u2705<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u672c\u4f8b\u4e2d\uff0c\u6211\u4eec\u4e3b\u8981\u4e86\u89e3Chroma\u7684\u4f7f\u7528\u3002<\/p>\n<h3><span class=\"ez-toc-section\" id=\"Chroma%E7%AE%80%E4%BB%8B\"><\/span>Chroma\u7b80\u4ecb<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>\u4ecb\u7ecd\uff1aChroma \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u5411\u91cf\u6570\u636e\u5e93\uff0c\u4e13\u4e3a\u5904\u7406\u548c\u5b58\u50a8\u9ad8\u7ef4\u5411\u91cf\u800c\u8bbe\u8ba1\uff0c\u7279\u522b\u9002\u7528\u4e8e\u673a\u5668\u5b66\u4e60\u548c\u6df1\u5ea6\u5b66\u4e60\u5e94\u7528\u3002<\/p>\n<p>\u7279\u6027\uff1a<\/p>\n<ul>\n<li><strong>\u9ad8\u6548\u7684\u5411\u91cf\u5b58\u50a8<\/strong>\uff1aChroma \u63d0\u4f9b\u9ad8\u6548\u7684\u5411\u91cf\u5b58\u50a8\u548c\u68c0\u7d22\u529f\u80fd\uff0c\u80fd\u591f\u5904\u7406\u5927\u89c4\u6a21\u6570\u636e\u96c6\u3002<\/li>\n<li><strong>\u76f8\u4f3c\u6027\u641c\u7d22<\/strong>\uff1a\u652f\u6301\u5feb\u901f\u7684\u76f8\u4f3c\u6027\u641c\u7d22\uff0c\u5141\u8bb8\u7528\u6237\u6839\u636e\u67e5\u8be2\u5411\u91cf\u627e\u5230\u6700\u76f8\u4f3c\u7684\u5411\u91cf\u3002<\/li>\n<li><strong>\u7075\u6d3b\u7684Embedding\u652f\u6301<\/strong>\uff1a\u53ef\u4ee5\u4e0e\u591a\u79cdEmbedding\u6a21\u578b\u96c6\u6210\u3002<\/li>\n<li><strong>\u6301\u4e45\u5316\u5b58\u50a8<\/strong>\uff1a\u652f\u6301\u5c06\u6570\u636e\u6301\u4e45\u5316\u5b58\u50a8\uff0c\u786e\u4fdd\u5728\u91cd\u542f\u540e\u6570\u636e\u4e0d\u4f1a\u4e22\u5931\u3002<\/li>\n<li><strong>\u5f00\u6e90\u548c\u793e\u533a\u652f\u6301<\/strong>\uff1aChroma \u662f\u5f00\u6e90\u9879\u76ee\uff0c\u7528\u6237\u53ef\u4ee5\u81ea\u7531\u4f7f\u7528\u548c\u4fee\u6539\uff0c\u540c\u65f6\u4e5f\u4eab\u53d7\u793e\u533a\u7684\u652f\u6301\u548c\u8d21\u732e\u3002<\/li>\n<\/ul>\n<p>\u8d44\u6599\uff1a<\/p>\n<ul>\n<li>\u5b98\u7f51\u4e3b\u9875\uff1a<a href=\"https:\/\/www.trychroma.com\/\">https:\/\/www.trychroma.com\/<\/a><\/li>\n<li>\u5b98\u7f51\u6587\u6863\uff1a<a href=\"https:\/\/docs.trychroma.com\/\">https:\/\/docs.trychroma.com\/<\/a><\/li>\n<li>Github\u4e3b\u9875\uff1a<a href=\"https:\/\/github.com\/chroma-core\/\">https:\/\/github.com\/chroma-core\/<\/a><\/li>\n<li>Github\u9879\u76ee\uff1a<a href=\"https:\/\/github.com\/chroma-core\/chroma\/\">https:\/\/github.com\/chroma-core\/chroma\/<\/a><\/li>\n<\/ul>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Chroma\u793a\u610f\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Chroma\u793a\u610f\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"Chroma%E4%BD%BF%E7%94%A8%E7%9A%84%E5%87%A0%E7%A7%8D%E6%96%B9%E5%BC%8F\"><\/span>Chroma\u4f7f\u7528\u7684\u51e0\u79cd\u65b9\u5f0f<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>\u521d\u6b21\u4e86\u89e3Chroma\u7684\u4f7f\u7528\u65f6\uff0c\u6211\u88ab\u4e94\u82b1\u516b\u95e8\u7684API\u641e\u5f97\u4e91\u91cc\u96fe\u91cc\uff0c\u4e00\u4f1a\u662fChromadb\uff0c\u4e00\u4f1a\u662flangchain_chroma&#8230;..<br \/>\n\u56e0\u6b64\uff0c\u67e5\u8be2\u4e86\u90e8\u5206\u8d44\u6599\u4e4b\u540e\uff0c\u603b\u7ed3Chroma\u7684\u4f7f\u7528\u65b9\u5f0f\u5927\u81f4\u662f\u4ee5\u4e0b\u4e09\u79cd\u65b9\u5f0f\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Chroma\u7684\u4f7f\u7528\u65b9\u5f0f.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Chroma\u7684\u4f7f\u7528\u65b9\u5f0f.png\" alt=\"\" \/><\/a><\/p>\n<ul>\n<li>\u7b2c\u4e00\u79cd\u65b9\u5f0f\uff1a\u662f\u4f7f\u7528Chromadb\u7684\u6e90\u751fAPI\u63a5\u53e3\u4f7f\u7528\uff0c\u5305\u62ec\u5e38\u89c1\u7684\u589e\u5220\u6539\u67e5\u63a5\u53e3\uff1a<code>create_collection()<\/code>\u3001<code>collection.add()<\/code>\u3001<code>collection.query()<\/code>&#8230;<\/li>\n<li>\u7b2c\u4e8c\u79cd\u65b9\u5f0f\uff1a\u662f\u901a\u8fc7Langchain\u7684\u7ec4\u5408\u5305\u4f7f\u7528Chroma\uff0c\u5e38\u89c1\u7684\u662f\uff1a<code>Chroma.from_documents()<\/code>\u3001<code>similarity_search()<\/code>\u3001<code>similarity_search_by_vector()<\/code>&#8230;<\/li>\n<li>\u7b2c\u4e09\u79cd\u65b9\u5f0f\uff1a\u662f\u901a\u8fc7Retriever\u68c0\u7d22\u5668\u7684\u65b9\u5f0f\u4f7f\u7528Chroma\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"%E9%80%9A%E8%BF%87Chromadb%E7%9A%84%E6%BA%90%E7%94%9F%E6%8E%A5%E5%8F%A3%E4%BD%BF%E7%94%A8Chroma\"><\/span>\u901a\u8fc7Chromadb\u7684\u6e90\u751f\u63a5\u53e3\u4f7f\u7528Chroma<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>Chroma \u53ef\u4ee5\u4ee5\u591a\u79cd\u5f62\u5f0f\u8fdb\u884c\u4f7f\u7528\uff1a<\/p>\n<ul>\n<li>\u7b2c\u4e00\u79cd\uff1ain-memory with ephemeral\u3002\u77ed\u6682\u6a21\u5f0f\uff0c\u6570\u636e\u4fdd\u5b58\u5728\u5185\u5b58\u4e2d\uff0c\u7a0b\u5e8f\u4e00\u65e6\u7ed3\u675f\u6570\u636e\u5c31\u88ab\u9500\u6bc1\u3002<\/li>\n<li>\u7b2c\u4e8c\u79cd\uff1ain-memory with persistance\u3002\u6301\u4e45\u5316\u6a21\u5f0f\uff0c\u6570\u636e\u4fdd\u5b58\u5728sqlite\u6570\u636e\u5e93\u4e2d\uff0c\u7a0b\u5e8f\u7ed3\u675f\u6570\u636e\u4ecd\u7136\u5b58\u5728\u3002<\/li>\n<li>\u7b2c\u4e09\u79cd\uff1aclient\/server\u6a21\u5f0f\u3002<\/li>\n<\/ul>\n<p>\u672c\u6b21\u5b9e\u8df5\u4e2d\uff0c\u4e3b\u8981\u5c1d\u8bd5 <code>\u6301\u4e45\u5316\u6a21\u5f0f<\/code> \u548c <code>Client\/Server\u6a21\u5f0f<\/code>\u3002<\/p>\n<h5><span class=\"ez-toc-section\" id=\"%E6%8C%81%E4%B9%85%E5%8C%96%E6%A8%A1%E5%BC%8F\"><\/span>\u6301\u4e45\u5316\u6a21\u5f0f<span class=\"ez-toc-section-end\"><\/span><\/h5>\n<p><strong>\u7b2c\u4e00\u6b65<\/strong>\uff1a\u5b89\u88c5chroma\u5e93<\/p>\n<pre><code class=\"language-bash\">pip install chromadb -i https:\/\/pypi.tuna.tsinghua.edu.cn\/simple<\/code><\/pre>\n<p><strong>\u7b2c\u4e8c\u6b65<\/strong>\uff1a\u521b\u5efaChromadb\u5b9e\u4f8b<\/p>\n<pre><code class=\"language-python\">import chromadb\nfrom chromadb import Settings\n\nchroma_client = chromadb.PersistentClient(path=&quot;chroma_db&quot;)\n<\/code><\/pre>\n<p><strong>\u7b2c\u4e09\u6b65<\/strong>\uff1a\u521b\u5efa\u6216\u83b7\u53d6\u88ab\u67e5\u8be2\u96c6\u5408<\/p>\n<pre><code class=\"language-python\"># \u65b9\u5f0f\u4e00\uff1a\u521b\u5efa\u4e00\u4e2a\u65b0\u7684\u96c6\u5408\n# collection = chroma_client.create_collection(name=&quot;my_collection&quot;)\n\n# \u65b9\u5f0f\u4e8c\uff1a\u5bf9\u4e8e\u5df2\u5b58\u5728\u7684\u96c6\u5408\uff0c\u53ef\u4ee5\u76f4\u63a5\u901a\u8fc7\u540d\u79f0\u83b7\u53d6\uff0c\u5982\u679c\u96c6\u5408\u4e0d\u5b58\u5728\u4f1a\u62a5\u9519\u3002\n# collection = chroma_client.get_collection(name=&quot;my_collection&quot;)\n\n# \u65b9\u5f0f\u4e09\uff1a\u5982\u679c\u7ed9\u5b9a\u540d\u79f0\u7684\u96c6\u5408\u5df2\u7ecf\u5b58\u5728\u5219\u76f4\u63a5\u83b7\u53d6\uff0c\u5426\u5219\u521b\u5efa\u3002\ncollection = chroma_client.get_or_create_collection(name=&quot;my_collection&quot;)\n<\/code><\/pre>\n<p><strong>\u7b2c\u56db\u6b65<\/strong>\uff1a\u6dfb\u52a0\u6570\u636e\u5230\u96c6\u5408\u4e2d<\/p>\n<pre><code class=\"language-python\">collection.add(\n    ids=[&quot;id1&quot;, &quot;id2&quot;, &quot;id3&quot;, &quot;id4&quot;, &quot;id5&quot;],\n    documents=[&quot;\u6d59\u6c5f\u7684\u7701\u4f1a\u662f\u676d\u5dde&quot;, &quot;\u6cb3\u5317\u7684\u7701\u4f1a\u662f\u77f3\u5bb6\u5e84&quot;, &quot;\u5c71\u4e1c\u7684\u7701\u4f1a\u662f\u6d4e\u5357&quot;, &quot;\u676d\u5dde\u662f\u4e2a\u7f8e\u4e3d\u7684\u57ce\u5e02&quot;, &quot;\u676d\u5dde\u4f4d\u4e8e\u6d59\u6c5f\u7701&quot;],\n    metadatas=[{&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}],\n)\n<\/code><\/pre>\n<p><strong>\u7b2c\u4e94\u6b65<\/strong>\uff1a\u67e5\u8be2\u96c6\u5408\u4e2d\u7684\u6570\u636e<\/p>\n<pre><code class=\"language-python\">\nresults = collection.query(\n    query_texts=[&quot;\u897f\u6e56\u5728\u54ea\u4e2a\u7701\uff1f&quot;],\n    n_results=2,\n)<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>query_texts: \u67e5\u8be2\u7684\u6587\u672c<\/li>\n<li>n_results: \u67e5\u8be2\u7ed3\u679c\u7684\u6570\u91cf<\/li>\n<\/ul>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">{&#039;ids&#039;: [[&#039;id5&#039;, &#039;id1&#039;]],\n &#039;distances&#039;: [[0.669669112082433, 0.6793151914879495]],\n &#039;metadatas&#039;: [[{&#039;source&#039;: &#039;myDoc&#039;}, {&#039;source&#039;: &#039;myDoc&#039;}]],\n &#039;embeddings&#039;: None,\n &#039;documents&#039;: [[&#039;\u676d\u5dde\u4f4d\u4e8e\u6d59\u6c5f\u7701&#039;, &#039;\u6d59\u6c5f\u7684\u7701\u4f1a\u662f\u676d\u5dde&#039;]],\n &#039;uris&#039;: None,\n &#039;data&#039;: None,\n &#039;included&#039;: [&#039;metadatas&#039;, &#039;documents&#039;, &#039;distances&#039;]}<\/code><\/pre>\n<h5><span class=\"ez-toc-section\" id=\"ClientServer%E6%A8%A1%E5%BC%8F\"><\/span>Client\/Server\u6a21\u5f0f<span class=\"ez-toc-section-end\"><\/span><\/h5>\n<p><strong>\u7b2c\u4e00\u6b65<\/strong>\uff1a\u542f\u52a8Chroma\u670d\u52a1\u7aef<\/p>\n<pre><code class=\"language-bash\">chroma run --path my_chroma<\/code><\/pre>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>&#8211;path\uff1a\u6307\u5b9aChroma\u670d\u52a1\u7aef\u5b58\u50a8\u6570\u636e\u7684\u8def\u5f84\u3002<\/li>\n<li>&#8211;host\uff1a\u53ef\u4ee5\u6307\u5b9aChroma\u670d\u52a1\u7aef\u76d1\u542c\u7684IP\u5730\u5740\uff0c\u9ed8\u8ba4localhost<\/li>\n<li>&#8211;port\uff1a\u53ef\u4ee5\u6307\u5b9aChroma\u670d\u52a1\u7aef\u76d1\u542c\u7684\u7aef\u53e3\u53f7\uff0c\u9ed8\u8ba48000<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u542f\u52a8chroma\u670d\u52a1\u5668.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u542f\u52a8chroma\u670d\u52a1\u5668.png\" alt=\"\" \/><\/a><\/li>\n<\/ul>\n<p><strong>\u7b2c\u4e8c\u6b65<\/strong>\uff1a\u521b\u5efaChromadb\u5b9e\u4f8b<\/p>\n<pre><code class=\"language-python\">import chromadb\nfrom chromadb import Client\nfrom chromadb import Settings\n\n# \u914d\u7f6e\u8fde\u63a5\u4fe1\u606f\nsetting = Settings(chroma_server_host=&quot;localhost&quot;, \n                  chroma_server_http_port=8000)\n\nchroma_client = Client(settings=setting)          <\/code><\/pre>\n<p><strong>\u7b2c\u4e09\u6b65<\/strong>\uff1a\u521b\u5efa\u6216\u83b7\u53d6\u88ab\u67e5\u8be2\u96c6\u5408<\/p>\n<pre><code class=\"language-python\">collection = chroma_client.get_or_create_collection(name=&quot;my_collection&quot;)<\/code><\/pre>\n<p><strong>\u7b2c\u56db\u6b65<\/strong>\uff1a\u6dfb\u52a0\u6570\u636e\u5230\u96c6\u5408\u4e2d<\/p>\n<pre><code class=\"language-python\">collection.add(\n    ids=[&quot;id1&quot;, &quot;id2&quot;, &quot;id3&quot;, &quot;id4&quot;, &quot;id5&quot;],\n    documents=[&quot;\u6d59\u6c5f\u7684\u7701\u4f1a\u662f\u676d\u5dde&quot;, &quot;\u6cb3\u5317\u7684\u7701\u4f1a\u662f\u77f3\u5bb6\u5e84&quot;, &quot;\u5c71\u4e1c\u7684\u7701\u4f1a\u662f\u6d4e\u5357&quot;, &quot;\u676d\u5dde\u662f\u4e2a\u7f8e\u4e3d\u7684\u57ce\u5e02&quot;, &quot;\u676d\u5dde\u4f4d\u4e8e\u6d59\u6c5f\u7701&quot;],\n    metadatas=[{&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}, {&quot;source&quot;: &quot;myDoc&quot;}],\n)<\/code><\/pre>\n<p><strong>\u7b2c\u4e94\u6b65<\/strong>\uff1a\u67e5\u8be2\u96c6\u5408\u4e2d\u7684\u6570\u636e<\/p>\n<pre><code class=\"language-python\">results = collection.query(\n    query_texts=[&quot;\u897f\u6e56\u5728\u54ea\u4e2a\u7701\uff1f&quot;],\n    n_results=2,\n)\nresults<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<\/p>\n<pre><code class=\"language-bash\">{&#039;ids&#039;: [[&#039;id5&#039;, &#039;id1&#039;]],\n &#039;distances&#039;: [[0.6696690917015076, 0.6793153882026672]],\n &#039;metadatas&#039;: [[{&#039;source&#039;: &#039;myDoc&#039;}, {&#039;source&#039;: &#039;myDoc&#039;}]],\n &#039;embeddings&#039;: None,\n &#039;documents&#039;: [[&#039;\u676d\u5dde\u4f4d\u4e8e\u6d59\u6c5f\u7701&#039;, &#039;\u6d59\u6c5f\u7684\u7701\u4f1a\u662f\u676d\u5dde&#039;]],\n &#039;uris&#039;: None,\n &#039;data&#039;: None,\n &#039;included&#039;: [&#039;metadatas&#039;, &#039;documents&#039;, &#039;distances&#039;]}<\/code><\/pre>\n<p>\u5bf9\u6bd4 \u672c\u5730\u6301\u4e45\u5316 \u548c Client\/Server\u6a21\u5f0f\uff0c\u4e24\u8005\u7684\u4f7f\u7528\u8fc7\u7a0b\u57fa\u672c\u4e00\u81f4\uff0c\u53ea\u662f\u5728Client\u5b9e\u4f8b\u5316\u65f6\u7565\u6709\u4e0d\u540c\u3002<\/p>\n<ul>\n<li>\u672c\u5730\u6301\u4e45\u5316\u6a21\u578b\uff1a<code>chroma_client = chromadb.PersistentClient(path=&quot;chroma_db&quot;)<\/code><\/li>\n<li>Client\/Server\u6a21\u5f0f\uff1a\u9996\u5148\u914d\u7f6e <code>Settings<\/code> \u8fde\u63a5\u4fe1\u606f\uff0c\u7136\u540e\u8c03\u7528 <code>chroma_client = Client(settings=setting)<\/code><\/li>\n<\/ul>\n<p>\u67e5\u770bchromadb\u7684\u6e90\u7801\uff0c\u53ef\u4ee5\u770b\u5230\u5176\u5185\u90e8\u5b9e\u73b0\u4e86DB\u5e38\u89c1\u7684\u5404\u7c7b\u589e\u5220\u6539\u67e5\u64cd\u4f5c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/chromadb\u6e90\u7801.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/chromadb\u6e90\u7801.png\" alt=\"\" \/><\/a><\/p>\n<blockquote>\n<p>\u5907\u6ce8\uff1a\u5b9e\u9645\u9879\u76ee\u4ea7\u54c1\u5f00\u53d1\u4e2d\uff0c\u77e5\u8bc6\u5e93\u9700\u8981\u901a\u8fc7\u4e0a\u8ff0\u7684chromadb\u5b9e\u73b0\u77e5\u8bc6\u5e93\u7684\u7ba1\u7406\uff08\u589e\u5220\u6539\u67e5\uff09\u3002<\/p>\n<\/blockquote>\n<h4><span class=\"ez-toc-section\" id=\"%E9%80%9A%E8%BF%87Langchain%E7%9A%84%E7%BB%84%E5%90%88%E5%8C%85%E4%BD%BF%E7%94%A8Chroma\"><\/span>\u901a\u8fc7Langchain\u7684\u7ec4\u5408\u5305\u4f7f\u7528Chroma<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u9664\u4e86\u4e0a\u8ff0<code>chromadb<\/code>\u7684\u65b9\u5f0f\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528<code>Langchain_chroma<\/code>\u7684\u65b9\u5f0f\u4f7f\u7528Chroma\uff0c\u4e24\u8005\u7684\u4f7f\u7528\u573a\u666f\u533a\u522b\u662f\uff1a<\/p>\n<p><strong>chromadb<\/strong>\uff1a<\/p>\n<ul>\n<li>\u81ea\u5b9a\u4e49\u5e94\u7528\uff1a\u5f53\u5f00\u53d1\u8005\u9700\u8981\u6784\u5efa\u7279\u5b9a\u7684\u5e94\u7528\uff0c\u4e14\u5bf9\u6570\u636e\u5e93\u7684\u64cd\u4f5c\u6709\u8f83\u9ad8\u7684\u7075\u6d3b\u6027\u9700\u6c42\u65f6\uff0c\u53ef\u4ee5\u4f7f\u7528 <code>chromadb<\/code>\u3002<\/li>\n<li>\u6570\u636e\u5206\u6790\uff1a\u9700\u8981\u5bf9\u5b58\u50a8\u7684\u6570\u636e\u8fdb\u884c\u590d\u6742\u67e5\u8be2\u548c\u5206\u6790\u65f6\uff0c\u4f7f\u7528\u5e95\u5c42\u63a5\u53e3\u53ef\u4ee5\u66f4\u597d\u5730\u6ee1\u8db3\u9700\u6c42\u3002<\/li>\n<li>\u9ad8\u6027\u80fd\u9700\u6c42\uff1a\u5728\u9700\u8981\u4f18\u5316\u6027\u80fd\u7684\u60c5\u51b5\u4e0b\uff0c\u5f00\u53d1\u8005\u53ef\u4ee5\u901a\u8fc7\u5e95\u5c42\u63a5\u53e3\u8fdb\u884c\u66f4\u7ec6\u81f4\u7684\u8c03\u6574\u3002<\/li>\n<\/ul>\n<p><strong>langchain_chroma<\/strong>\uff1a<\/p>\n<ul>\n<li>\u5feb\u901f\u539f\u578b\u5f00\u53d1\uff1a\u5f53\u5f00\u53d1\u8005\u5e0c\u671b\u5feb\u901f\u6784\u5efa\u539f\u578b\u6216 MVP\uff08\u6700\u5c0f\u53ef\u884c\u4ea7\u54c1\uff09\u65f6\uff0c<code>langchain_chroma <\/code>\u63d0\u4f9b\u4e86\u4fbf\u6377\u7684\u63a5\u53e3\u3002<\/li>\n<li>NLP\u68c0\u7d22\uff1a\u63d0\u4f9b\u66f4\u9ad8\u5c42\u6b21\u7684\u62bd\u8c61\uff0c\u9002\u7528\u4e8e\u68c0\u7d22\u7c7b\u4efb\u52a1\uff0c\u5982\u6587\u6863\u68c0\u7d22\u3001\u8bed\u4e49\u641c\u7d22\u7b49\u3002<\/li>\n<\/ul>\n<p>\u5177\u4f53\u4f7f\u7528\u65b9\u6cd5\uff1a<\/p>\n<pre><code class=\"language-python\">from langchain.text_splitter import CharacterTextSplitter\nfrom langchain_chroma import Chroma\nfrom langchain.document_loaders import PyMuPDFLoader\nfrom chromadb import Settings\nfrom utils import get_ernie_models\nfrom utils import get_qwen_models\n\n# \u8fde\u63a5\u5927\u6a21\u578b\nllm_qwen, chat_qwen, embed_qwen = get_qwen_models()\nllm_ernie, chat_ernie, embed_ernie = get_ernie_models()\n\n# \u52a0\u8f7d\u6587\u6863\npdf_loader = PyMuPDFLoader(&quot;testfiles\/\u897f\u6e38\u8bb0.pdf&quot;)\ndocuments = pdf_loader.load()\n\n# \u5207\u5206\u6587\u6863\ntext_splitter = CharacterTextSplitter(chunk_size=100, chunk_overlap=32)\ndocs = text_splitter.split_documents(documents)\n\n# \u914d\u7f6e\u8fde\u63a5\u4fe1\u606f\n\nclient = chromadb.HttpClient(host=&#039;localhost&#039;, port=8000)\n\nchroma_db = Chroma(\n    embedding_function=embed_qwen,\n    client=client\n)\n\nbatch_size = 6  # \u6bcf\u6b21\u5904\u7406\u7684\u6837\u672c\u6570\u91cf\n\n# \u5206\u6279\u5165\u5e93\nfor i in range(0, len(docs), batch_size):\n    batch = docs[i:i + batch_size]  # \u83b7\u53d6\u5f53\u524d\u6279\u6b21\u7684\u6837\u672c\n    chroma_db.add_documents(documents=batch)  # \u5165\u5e93\n\n# \u67e5\u8be2\nquery = &quot;\u767d\u9aa8\u7cbe\u88ab\u6253\u6b7b\u51e0\u6b21\uff1f&quot;\ndocs = chroma_db.similarity_search(query, k=3)\n\nprint(len(docs))\n\n# \u6253\u5370\u7ed3\u679c\nfor doc in docs:\n    print(&quot;=&quot;*100)\n    print(doc.page_content)\n<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u4e09\u6253\u767d\u9aa8\u7cbe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u4e09\u6253\u767d\u9aa8\u7cbe.png\" alt=\"\" \/><\/a><\/p>\n<p>\u8bf4\u660e\uff1a<\/p>\n<ul>\n<li>\u5728langchain_chroma\u4e2d\uff0c\u901a\u8fc7client_settings\u53c2\u6570\uff0c\u53ef\u4ee5l\u8fde\u63a5Chroma\u670d\u52a1\u7aef\u3002<\/li>\n<li>\u5728Chroma\u4e2d\uff0c\u901a\u8fc7add_documents\u65b9\u6cd5\uff0c\u53ef\u4ee5\u6279\u91cf\u5165\u5e93\uff1b\u4e0d\u8fc7\u5bf9\u4e8e\u897f\u6e38\u8bb0\u8fd9\u6837\u7684\u5c0f\u8bf4\u6765\u8bf4\uff0c\u5165\u5e93\u65f6\u95f4\u6bd4\u8f83\u957f\uff0c\u7ea65\u5206\u949f\u3002<\/li>\n<\/ul>\n<blockquote>\n<p>\u897f\u6e38\u8bb0\u53ef\u4ee5\u4ece<a href=\"https:\/\/pan.quark.cn\/s\/93b12d4185eb#\/list\/share\/e912eb82cdc34d2fb893e62a880ebd9b-%E8%A5%BF%E6%B8%B8%E8%AE%B0(%E8%AE%B0%E5%BE%97%E4%BF%9D%E5%AD%98%E5%88%B0%E4%BA%91%E7%9B%98)\">\u5938\u514b\u7f51\u76d8\uff1a\u897f\u6e38\u8bb0<\/a>\u4e0b\u8f7d<\/p>\n<\/blockquote>\n<h4><span class=\"ez-toc-section\" id=\"%E9%80%9A%E8%BF%87Retriever%E6%A3%80%E7%B4%A2%E5%99%A8%E7%9A%84%E6%96%B9%E5%BC%8F%E4%BD%BF%E7%94%A8Chroma\"><\/span>\u901a\u8fc7Retriever\u68c0\u7d22\u5668\u7684\u65b9\u5f0f\u4f7f\u7528Chroma<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u5728\u5b9e\u9645\u9879\u76ee\u5f00\u53d1\u4e2d\uff0c\u4e00\u822c\u9700\u8981\u5c06Chroma\u7684\u5411\u91cf\u5e93\u4f5c\u4e3a\u68c0\u7d22\u5668\uff08Retriever\uff09\u4e0e\u6a21\u578b\u7ec4\u6210chain\u94fe\uff0c\u4ee5\u5b9e\u73b0\u95ee\u7b54\u7cfb\u7edf\u7684\u95ee\u7b54\u80fd\u529b\u3002<\/p>\n<h5><span class=\"ez-toc-section\" id=\"%E4%BD%BF%E7%94%A8%E6%96%B9%E6%B3%95\"><\/span>\u4f7f\u7528\u65b9\u6cd5<span class=\"ez-toc-section-end\"><\/span><\/h5>\n<p>\u5728\u4e0a\u8ff0Langchain_chroma\u7684\u4ee3\u7801\u57fa\u7840\u4e0a\uff0c\u589e\u52a0\u4ee5\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre><code class=\"language-python\">retriever = chroma_db.as_retriever(search_type=&quot;similarity_score_threshold&quot;,\n                              search_kwargs={&quot;k&quot;: 4, &quot;score_threshold&quot;: 0.1})\n\nretriever.invoke(input=&quot;\u767d\u9aa8\u7cbe\u88ab\u6253\u6b7b\u51e0\u6b21\uff1f&quot;)<\/code><\/pre>\n<p>\u8fd0\u884c\u7ed3\u679c\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Retriever\u8fd0\u884c\u65b9\u5f0f.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Retriever\u8fd0\u884c\u65b9\u5f0f.png\" alt=\"\" \/><\/a><\/p>\n<h5><span class=\"ez-toc-section\" id=\"%E5%87%BD%E6%95%B0%E8%AF%B4%E6%98%8E%EF%BC%9A\"><\/span>\u51fd\u6570\u8bf4\u660e\uff1a<span class=\"ez-toc-section-end\"><\/span><\/h5>\n<p>1\u3001as_retriever\uff1a<br \/>\n<strong>\u529f\u80fd<\/strong>\uff1aas_retriever \u65b9\u6cd5\u5c06 Chroma \u6570\u636e\u5e93\u8f6c\u6362\u4e3a\u4e00\u4e2a\u68c0\u7d22\u5668\uff08retriever\uff09\uff0c\u4f7f\u5176\u80fd\u591f\u6839\u636e\u7ed9\u5b9a\u7684\u67e5\u8be2\u8fdb\u884c\u6587\u6863\u68c0\u7d22\u3002\u8fd9\u4e2a\u68c0\u7d22\u5668\u53ef\u4ee5\u4f7f\u7528\u4e0d\u540c\u7684\u641c\u7d22\u7c7b\u578b\u548c\u53c2\u6570\u6765\u4f18\u5316\u68c0\u7d22\u7ed3\u679c\u3002<\/p>\n<p><strong>\u53c2\u6570<\/strong><\/p>\n<ul>\n<li><code>search_type<\/code>: \u6307\u5b9a\u68c0\u7d22\u7684\u7c7b\u578b\u3002\u5e38\u89c1\u7684\u9009\u9879\u5305\u62ec\uff1a\n<ul>\n<li><code>&quot;similarity_score_threshold&quot;<\/code>\uff1a\u6839\u636e\u76f8\u4f3c\u5ea6\u5206\u6570\u8fdb\u884c\u68c0\u7d22\uff0c\u53ea\u6709\u8d85\u8fc7\u6307\u5b9a\u9608\u503c\u7684\u7ed3\u679c\u624d\u4f1a\u88ab\u8fd4\u56de\u3002<\/li>\n<\/ul>\n<\/li>\n<li><code>search_kwargs<\/code>: \u5305\u542b\u4e0e\u641c\u7d22\u76f8\u5173\u7684\u989d\u5916\u53c2\u6570\u3002\u5e38\u7528\u53c2\u6570\u5305\u62ec\uff1a\n<ul>\n<li><code>k<\/code>: \u6307\u5b9a\u8981\u8fd4\u56de\u7684\u6700\u76f8\u5173\u6587\u6863\u7684\u6570\u91cf\u3002\u4f8b\u5982\uff0ck=4 \u8868\u793a\u8fd4\u56de\u524d 4 \u4e2a\u76f8\u5173\u6587\u6863\u3002<\/li>\n<li><code>score_threshold<\/code>:\u8bbe\u7f6e\u76f8\u4f3c\u5ea6\u5206\u6570\u7684\u9608\u503c\u3002\u53ea\u6709\u76f8\u4f3c\u5ea6\u5206\u6570\u9ad8\u4e8e\u8be5\u9608\u503c\u7684\u6587\u6863\u624d\u4f1a\u88ab\u8fd4\u56de\u3002\u4f8b\u5982\uff0cscore_threshold=0.1 \u8868\u793a\u8fd4\u56de\u5206\u6570\u5927\u4e8e 0.1 \u7684\u6587\u6863\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>2\u3001invoke\uff1a<br \/>\n<strong>\u529f\u80fd<\/strong>\uff1ainvoke \u65b9\u6cd5\u7528\u4e8e\u6267\u884c\u5b9e\u9645\u7684\u68c0\u7d22\u64cd\u4f5c\uff0c\u6839\u636e\u63d0\u4f9b\u7684\u8f93\u5165\u67e5\u8be2\u8fd4\u56de\u76f8\u5173\u7684\u6587\u6863\u3002<br \/>\n<strong>\u53c2\u6570<\/strong>\uff1a<\/p>\n<ul>\n<li><code>input<\/code>:\u8981\u67e5\u8be2\u7684\u8f93\u5165\u6587\u672c\u3002<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\"><\/span>\u5185\u5bb9\u5c0f\u7ed3<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<ul>\n<li>RAG\u7684\u5efa\u5e93\u7684\u6574\u4f53\u6d41\u7a0b\u4e3a\uff1a\u6587\u6863\u8bfb\u53d6(LOAD) -&gt; \u6587\u6863\u5207\u5206(SPLIT) -&gt; \u5411\u91cf\u5316(EMBED) -&gt; \u5b58\u50a8(STORE)<\/li>\n<li>\u5728\u6587\u6863\u8bfb\u53d6(LOAD)\u65f6\uff1a\n<ul>\n<li>langchain_community.document_loaders\u6709\u591a\u4e2a\u52a0\u8f7d\u5668\uff0c\u53ef\u4ee5\u52a0\u8f7d\u591a\u79cd\u683c\u5f0f\u7684\u6587\u4ef6\uff0c\u5982PDF\u3001EXCEL\u3001PPT\u3001WORD\u3001MARKDOWN\u3001HTML\u7b49\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u5728\u6587\u6863\u5207\u5206(SPLIT)\u65f6\uff1a\n<ul>\n<li>\u6709\u5b57\u7b26\u7ea7\u5207\u5206\u3001\u9012\u5f52\u65b9\u5f0f\u5207\u5206\u3001Token\u65b9\u5f0f\u5207\u5206\u7b49\uff0c\u4e00\u822c\u60c5\u51b5\u53ef\u4ee5\u4f7f\u7528\u9012\u5f52RecursiveCharacterTextSplitter\u5207\u5206\u3002<\/li>\n<li>\u4f7f\u7528TokenTextSplitter\u5207\u5206\u65f6\uff0c\u6ce8\u610f\u9009\u62e9\u5bf9\u5e94\u7684encoding_name\uff0c\u5982\uff1a&#8217;o200k_base&#8217;\u3002<\/li>\n<li>\u5bf9\u4e8eMarkdown\u6587\u6863\uff0c\u53ef\u4ee5\u4f7f\u7528MarkdownTextSplitter\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u5728\u5411\u91cf\u5316(EMBED)\u65f6\uff1a\n<ul>\n<li>\u6709\u5728\u7ebf\u5316\u548c\u672c\u5730\u5316\uff0c\u5bf9\u4e8e\u6570\u636e\u654f\u611f\u6709\u672c\u5730\u5316\u90e8\u7f72\u7684\u60c5\u51b5\u4e0b\uff0c\u9700\u8981\u8003\u8651\u4f7f\u7528\u672c\u5730\u6a21\u578b\u8fdb\u884c\u5411\u91cf\u5316\u7684\u65b9\u6848\u3002<\/li>\n<li>\u5728\u7ebf\u7684\u5411\u91cf\u5316\u6709HuggingFaceEmbeddings\u548c\u7b2c\u4e09\u65b9\u6a21\u578b(\u5982\uff1aQianfanEmbeddings)<\/li>\n<li>\u5728\u6280\u672f\u539f\u7406\u5c42\u9762\uff0c\u8ba1\u7b97<code>\u77e5\u8bc6<\/code>\u4e0e<code>\u67e5\u8be2<\/code>\u76f8\u4f3c\u5ea6\u4e00\u822c\u4f7f\u7528\u7684\u662f\u6b27\u5f0f\u8ddd\u79bb\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u5728\u5b58\u50a8(STORE)\u65f6\uff1a\n<ul>\n<li>\u53ef\u4f9b\u9009\u62e9\u7684\u5411\u91cf\u6570\u636e\u5e93\u6709\u591a\u79cd\uff0c\u5176\u4e2d\u8f83\u4e3a\u5e38\u7528\u7684\u662fChroma\u3002<\/li>\n<li>Chroma\u652f\u6301\u591a\u79cd\u8fd0\u884c\u65b9\u5f0f\uff1a\u77ed\u6682\u6a21\u5f0f\u3001\u6301\u4e45\u5316\u6a21\u5f0f\u548cClient\/Server\u6a21\u5f0f\u3002<\/li>\n<li>\u4f7f\u7528chromadb\u7684\u539f\u751f\u63a5\u53e3\uff0c\u53ef\u4ee5\u8fdb\u884c\u589e\u5220\u6539\u67e5\u7b49\u64cd\u4f5c\uff0c\u9002\u7528\u4e8e\u77e5\u8bc6\u7684\u7ba1\u7406\u3002<\/li>\n<li>\u4f7f\u7528langchain_chroma\uff0c\u53ef\u4ee5\u5feb\u901f\u539f\u578b\u5f00\u53d1\uff0c\u9002\u7528\u4e8e\u68c0\u7d22\u7c7b\u4efb\u52a1\u3002<\/li>\n<li>\u4f7f\u7528retriever\uff0c\u53ef\u4ee5\u5c06Chroma\u7684\u5411\u91cf\u5e93\u4f5c\u4e3a\u68c0\u7d22\u5668\uff08Retriever\uff09\uff0c\u4e0e\u6a21\u578b\u7ec4\u6210chain\u94fe\uff0c\u5b9e\u73b0\u95ee\u7b54\u7cfb\u7edf\u7684\u95ee\u7b54\u80fd\u529b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\"><\/span>\u53c2\u8003\u8d44\u6599<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><a href=\"https:\/\/api.python.langchain.com\/en\/latest\/community_api_reference.html#module-langchain_community.document_loaders\">Langchain_community API\u8bf4\u660e\u6587\u6863<\/a><\/p>\n<p><a href=\"https:\/\/zhuanlan.zhihu.com\/p\/665715823\">\u77e5\u4e4e\uff1a\u5411\u91cf\u6570\u636e\u5e93Chroma\u6781\u7b80\u6559\u7a0b<\/a><\/p>\n<p><a href=\"https:\/\/cloud.tencent.com\/developer\/article\/2312534\">\u5411\u91cf\u6570\u636e\u5e93\uff5c\u4e00\u6587\u5168\u9762\u4e86\u89e3\u5411\u91cf\u6570\u636e\u5e93\u7684\u57fa\u672c\u6982\u5ff5\u3001\u539f\u7406\u3001\u7b97\u6cd5\u3001\u9009\u578b<\/a><\/p>\n<p><a href=\"https:\/\/blog.csdn.net\/engchina\/article\/details\/131868860\">CSDN:Langchain \u548c Chroma \u7684\u96c6\u6210<\/a><\/p>\n<p><a href=\"https:\/\/blog.csdn.net\/HRG520JN\/article\/details\/137857134\">CSDN:LangChain\u6559\u7a0b | Retrival\u4e4bRetrievers\u8be6\u89e3 | \u68c0\u7d22\u5668\u6559\u7a0b<\/a><\/p>\n<p align=\"center\">\u6b22\u8fce\u5173\u6ce8\u516c\u4f17\u53f7\u4ee5\u83b7\u5f97\u6700\u65b0\u7684\u6587\u7ae0\u548c\u65b0\u95fb<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" alt=\"\" \/><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u524d\u8a00 \u5728\u4e0a\u4e00\u7ae0\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day27\uff1a\u5927 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":15979,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"aside","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"default","adv-header-id-meta":"","stick-header-meta":"default","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[28],"tags":[71,68],"class_list":["post-18186","post","type-post","status-publish","format-aside","has-post-thumbnail","hentry","category-blog","tag-rag","tag-68","post_format-post-format-aside"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.9 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/17aitech.com\/?p=18186\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186\"},\"author\":{\"name\":\"Dongming\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/#\\\/schema\\\/person\\\/3d23bb6f7f115fcefc9ae7803a691739\"},\"headline\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b\",\"datePublished\":\"2024-08-24T16:45:10+00:00\",\"dateModified\":\"2024-10-08T07:08:49+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186\"},\"wordCount\":428,\"commentCount\":0,\"image\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/rag\u7684\u6d41\u7a0b.png\",\"keywords\":[\"RAG\",\"\u5927\u6a21\u578b\"],\"articleSection\":[\"\u535a\u5ba2\u6587\u7ae0\"],\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/17aitech.com\\\/?p=18186#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186\",\"url\":\"https:\\\/\\\/17aitech.com\\\/?p=18186\",\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/rag\u7684\u6d41\u7a0b.png\",\"datePublished\":\"2024-08-24T16:45:10+00:00\",\"dateModified\":\"2024-10-08T07:08:49+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/#\\\/schema\\\/person\\\/3d23bb6f7f115fcefc9ae7803a691739\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/17aitech.com\\\/?p=18186\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#primaryimage\",\"url\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/rag\u7684\u6d41\u7a0b.png\",\"contentUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/rag\u7684\u6d41\u7a0b.png\",\"width\":2583,\"height\":1299},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=18186#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\\\/\\\/17aitech.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/#website\",\"url\":\"https:\\\/\\\/17aitech.com\\\/\",\"name\":\"\u4e00\u8d77AI\u6280\u672f\",\"description\":\"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca\",\"alternateName\":\"\u4e00\u8d77AI\u6280\u672f\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/17aitech.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/#\\\/schema\\\/person\\\/3d23bb6f7f115fcefc9ae7803a691739\",\"name\":\"Dongming\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"url\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"contentUrl\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"caption\":\"Dongming\"},\"description\":\"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002\",\"sameAs\":[\"http:\\\/\\\/17aitech.com\"],\"url\":\"https:\\\/\\\/17aitech.com\\\/?page_id=33738&user=1\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/17aitech.com\/?p=18186","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/17aitech.com\/?p=18186#article","isPartOf":{"@id":"https:\/\/17aitech.com\/?p=18186"},"author":{"name":"Dongming","@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739"},"headline":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b","datePublished":"2024-08-24T16:45:10+00:00","dateModified":"2024-10-08T07:08:49+00:00","mainEntityOfPage":{"@id":"https:\/\/17aitech.com\/?p=18186"},"wordCount":428,"commentCount":0,"image":{"@id":"https:\/\/17aitech.com\/?p=18186#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png","keywords":["RAG","\u5927\u6a21\u578b"],"articleSection":["\u535a\u5ba2\u6587\u7ae0"],"inLanguage":"zh-Hans","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/17aitech.com\/?p=18186#respond"]}]},{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=18186","url":"https:\/\/17aitech.com\/?p=18186","name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=18186#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=18186#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png","datePublished":"2024-08-24T16:45:10+00:00","dateModified":"2024-10-08T07:08:49+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=18186#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=18186"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=18186#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/rag\u7684\u6d41\u7a0b.png","width":2583,"height":1299},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=18186#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day28\uff1a\u5927\u6a21\u578b\u4e4b\u6df1\u5165\u63a2\u7d22RAG\u6d41\u7a0b"}]},{"@type":"WebSite","@id":"https:\/\/17aitech.com\/#website","url":"https:\/\/17aitech.com\/","name":"\u4e00\u8d77AI\u6280\u672f","description":"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca","alternateName":"\u4e00\u8d77AI\u6280\u672f","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/17aitech.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739","name":"Dongming","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","url":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","caption":"Dongming"},"description":"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002","sameAs":["http:\/\/17aitech.com"],"url":"https:\/\/17aitech.com\/?page_id=33738&user=1"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/18186","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=18186"}],"version-history":[{"count":14,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/18186\/revisions"}],"predecessor-version":[{"id":33216,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/18186\/revisions\/33216"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/media\/15979"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=18186"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=18186"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=18186"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}