{"id":1157,"date":"2024-04-22T08:32:18","date_gmt":"2024-04-22T00:32:18","guid":{"rendered":"https:\/\/www.nicekj.com\/?p=1157"},"modified":"2024-04-22T08:32:32","modified_gmt":"2024-04-22T00:32:32","slug":"gpt4allchatgptbendisiyouhuabushuzhongshengmianfei","status":"publish","type":"post","link":"https:\/\/www.nicekj.com\/gpt4allchatgptbendisiyouhuabushuzhongshengmianfei.html","title":{"rendered":"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39"},"content":{"rendered":"<p>\u5728\u672c\u6587\u4e2d\uff0c\u6211\u4eec\u5c06\u5b66\u4e60\u5982\u4f55\u5728\u4ec5\u4f7f\u7528CPU\u7684\u8ba1\u7b97\u673a\u4e0a\u90e8\u7f72\u548c\u4f7f\u7528GPT4All\u6a21\u578b\uff08\u6211\u6b63\u5728\u4f7f\u7528\u6ca1\u6709GPU\u7684Macbook Pro\uff01\uff09\u5e76\u5b66\u4e60\u5982\u4f55\u4f7f\u7528Python\u4e0e\u6211\u4eec\u7684\u6587\u6863\u8fdb\u884c\u4ea4\u4e92\u3002\u4e00\u7ec4PDF\u6587\u4ef6\u6216\u5728\u7ebf\u6587\u7ae0\u5c06\u6210\u4e3a\u6211\u4eec\u95ee\u7b54\u7684\u77e5\u8bc6\u5e93\u3002<\/p>\n<h1 data-id=\"heading-0\">GPT4All\u662f\u4ec0\u4e48<\/h1>\n<p>\u6839\u636e\u5b98\u65b9\u7f51\u7ad9GPT4All\u7684\u63cf\u8ff0\uff0c\u5b83\u662f\u4e00\u4e2a\u514d\u8d39\u4f7f\u7528\u3001\u672c\u5730\u8fd0\u884c\u7684\u3001\u6ce8\u91cd\u9690\u79c1\u7684\u804a\u5929\u673a\u5668\u4eba\u3002\u4e0d\u9700\u8981GPU\u6216\u4e92\u8054\u7f51\u3002<\/p>\n<p>GTP4All\u662f\u4e00\u4e2a\u751f\u6001\u7cfb\u7edf\uff0c\u7528\u4e8e\u8bad\u7ec3\u548c\u90e8\u7f72\u5728\u6d88\u8d39\u7ea7CPU\u4e0a\u672c\u5730\u8fd0\u884c\u7684\u5f3a\u5927\u4e14\u5b9a\u5236\u7684\u5927\u578b\u8bed\u8a00\u6a21\u578b\u3002<\/p>\n<p>\u6211\u4eec\u7684GPT4All\u6a21\u578b\u662f\u4e00\u4e2a4GB\u7684\u6587\u4ef6\uff0c\u60a8\u53ef\u4ee5\u4e0b\u8f7d\u5e76\u8fde\u63a5\u5230GPT4All\u5f00\u6e90\u751f\u6001\u7cfb\u7edf\u8f6f\u4ef6\u4e2d\u3002Nomic AI\u63d0\u4f9b\u9ad8\u8d28\u91cf\u548c\u5b89\u5168\u7684\u8f6f\u4ef6\u751f\u6001\u7cfb\u7edf\uff0c\u52aa\u529b\u5b9e\u73b0\u4e2a\u4eba\u548c\u7ec4\u7ec7\u8f7b\u677e\u8bad\u7ec3\u548c\u672c\u5730\u5b9e\u65bd\u81ea\u5df1\u7684\u5927\u578b\u8bed\u8a00\u6a21\u578b\u3002<\/p>\n<h1 data-id=\"heading-1\">\u5b83\u662f\u5982\u4f55\u5de5\u4f5c\u7684\uff1f<\/h1>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/474fbef7e5cbb8d66d1439465b59dd70.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u8be5\u8fc7\u7a0b\u975e\u5e38\u7b80\u5355\uff08\u5f53\u60a8\u4e86\u89e3\u4e4b\u540e\uff09\uff0c\u5e76\u4e14\u53ef\u4ee5\u5728\u5176\u4ed6\u6a21\u578b\u4e0a\u91cd\u590d\u3002\u5177\u4f53\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n<ul>\n<li>\n<p>\u52a0\u8f7dGPT4All\u6a21\u578b<\/p>\n<\/li>\n<li>\n<p>\u4f7f\u7528Langchain\u68c0\u7d22\u5e76\u52a0\u8f7d\u6211\u4eec\u7684\u6587\u6863<\/p>\n<\/li>\n<li>\n<p>\u5c06\u6587\u6863\u5206\u5272\u6210\u5c0f\u5757\uff0c\u4ee5\u4fbf\u5d4c\u5165\u5f0f\u80fd\u591f\u7406\u89e3<\/p>\n<\/li>\n<li>\n<p>\u4f7f\u7528FAISS\u6839\u636e\u6211\u4eec\u8981\u4f20\u9012\u7ed9GPT4All\u7684\u95ee\u9898\u521b\u5efa\u5d4c\u5165\u5f0f\u5411\u91cf\u6570\u636e\u5e93<\/p>\n<\/li>\n<li>\n<p>\u5728\u57fa\u4e8e\u95ee\u9898\u7684\u8bed\u5883\u4e2d\u5728\u5d4c\u5165\u5f0f\u5411\u91cf\u6570\u636e\u5e93\u4e0a\u6267\u884c\u76f8\u4f3c\u6027\u641c\u7d22\uff08\u8bed\u4e49\u641c\u7d22\uff09\uff1a\u8fd9\u5c06\u4f5c\u4e3a\u6211\u4eec\u95ee\u9898\u7684\u4e0a\u4e0b\u6587<\/p>\n<\/li>\n<li>\n<p>\u4f7f\u7528Langchain\u5c06\u95ee\u9898\u548c\u4e0a\u4e0b\u6587\u63d0\u4f9b\u7ed9GPT4All\uff0c\u5e76\u7b49\u5f85\u7b54\u6848\u3002<\/p>\n<\/li>\n<\/ul>\n<p>\u6240\u4ee5\u6211\u4eec\u9700\u8981\u7684\u662f\u5d4c\u5165\u5f0f\u5411\u91cf\u3002\u5d4c\u5165\u5f0f\u5411\u91cf\u662f\u4e00\u6761\u4fe1\u606f\u7684\u6570\u503c\u8868\u793a\uff0c\u4f8b\u5982\u6587\u672c\u3001\u6587\u6863\u3001\u56fe\u50cf\u3001\u97f3\u9891\u7b49\u3002\u8be5\u8868\u793a\u6355\u6349\u4e86\u88ab\u5d4c\u5165\u7684\u8bed\u4e49\u542b\u4e49\uff0c\u8fd9\u6b63\u662f\u6211\u4eec\u6240\u9700\u8981\u7684\u3002\u5bf9\u4e8e\u8fd9\u4e2a\u9879\u76ee\uff0c\u6211\u4eec\u65e0\u6cd5\u4f9d\u8d56\u4e8e\u91cd\u578bGPU\u6a21\u578b\uff1a\u56e0\u6b64\uff0c\u6211\u4eec\u5c06\u4e0b\u8f7dAlpaca\u539f\u751f\u6a21\u578b\uff0c\u5e76\u4f7f\u7528Langchain\u4e2d\u7684LlamaCppEmbeddings\u3002\u4e0d\u7528\u62c5\u5fc3\uff01\u6bcf\u4e00\u6b65\u90fd\u6709\u8be6\u7ec6\u7684\u89e3\u91ca\u3002<\/p>\n<h1 data-id=\"heading-2\">Let\u2019s start coding<\/h1>\n<p><strong>\u521b\u5efa\u865a\u62df\u73af\u5883<\/strong><\/p>\n<p>\u4e3a\u4f60\u7684\u65b0Python\u9879\u76ee\u521b\u5efa\u4e00\u4e2a\u65b0\u7684\u6587\u4ef6\u5939\uff0c\u4f8b\u5982GPT4ALL_Fabio\uff08\u8bf7\u7528\u4f60\u7684\u540d\u5b57\u66ff\u6362Fabio\uff09\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">mkdir GPT4ALL_Fabiocd GPT4ALL_Fabio<\/span>\n<\/code><\/pre>\n<p>\u63a5\u4e0b\u6765\uff0c\u521b\u5efa\u4e00\u4e2a\u65b0\u7684Python\u865a\u62df\u73af\u5883\u3002\u5982\u679c\u4f60\u5b89\u88c5\u4e86\u591a\u4e2aPython\u7248\u672c\uff0c\u8bf7\u6307\u5b9a\u4f60\u60f3\u8981\u4f7f\u7528\u7684\u7248\u672c\u3002\u5728\u8fd9\u4e2a\u4f8b\u5b50\u4e2d\uff0c\u6211\u5c06\u4f7f\u7528\u4e0ePython 3.10\u5173\u8054\u7684\u4e3b\u8981\u5b89\u88c5\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">python3 -m venv .venv<\/span>\n<\/code><\/pre>\n<p>\u547d\u4ee4python3 -m venv .venv\u521b\u5efa\u4e86\u4e00\u4e2a\u540d\u4e3a.venv\u7684\u65b0\u865a\u62df\u73af\u5883\uff08\u70b9\u53f7\u4f1a\u521b\u5efa\u4e00\u4e2a\u540d\u4e3avenv\u7684\u9690\u85cf\u76ee\u5f55\uff09\u3002<\/p>\n<p>\u865a\u62df\u73af\u5883\u63d0\u4f9b\u4e86\u4e00\u4e2a\u9694\u79bb\u7684Python\u5b89\u88c5\u73af\u5883\uff0c\u5141\u8bb8\u4f60\u4ec5\u9488\u5bf9\u7279\u5b9a\u9879\u76ee\u5b89\u88c5\u8f6f\u4ef6\u5305\u548c\u4f9d\u8d56\u9879\uff0c\u800c\u4e0d\u4f1a\u5f71\u54cd\u7cfb\u7edf\u8303\u56f4\u7684Python\u5b89\u88c5\u6216\u5176\u4ed6\u9879\u76ee\u3002\u8fd9\u79cd\u9694\u79bb\u6709\u52a9\u4e8e\u4fdd\u6301\u4e00\u81f4\u6027\uff0c\u5e76\u9632\u6b62\u4e0d\u540c\u9879\u76ee\u9700\u6c42\u4e4b\u95f4\u7684\u6f5c\u5728\u51b2\u7a81\u3002<\/p>\n<p>\u4e00\u65e6\u865a\u62df\u73af\u5883\u521b\u5efa\u597d\u4e86\uff0c\u4f60\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u6765\u6fc0\u6d3b\u5b83\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">source .venv\/<span class=\"hljs-built_in\">bin<\/span>\/activate<\/span>\n<\/code><\/pre>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/47f558e965637d2ca072dfd366a4e8c8.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p><strong>\u5b89\u88c5\u4f9d\u8d56<\/strong><\/p>\n<p>\u5bf9\u4e8e\u6211\u4eec\u6b63\u5728\u6784\u5efa\u7684\u9879\u76ee\uff0c\u6211\u4eec\u4e0d\u9700\u8981\u592a\u591a\u7684\u8f6f\u4ef6\u5305\u3002\u6211\u4eec\u53ea\u9700\u8981\u4ee5\u4e0b\u51e0\u4e2a\uff1a<\/p>\n<ul>\n<li>GPT4All\u7684Python\u7ed1\u5b9a<\/li>\n<li>Langchain\u7528\u4e8e\u4e0e\u6211\u4eec\u7684\u6587\u6863\u4ea4\u4e92<\/li>\n<\/ul>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">pip install pygpt4all==<span class=\"hljs-number\">1.0<\/span><span class=\"hljs-number\">.1<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">pip install pyllamacpp==<span class=\"hljs-number\">1.0<\/span><span class=\"hljs-number\">.6<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">pip install langchain==<span class=\"hljs-number\">0.0<\/span><span class=\"hljs-number\">.149<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\">pip install unstructured==<span class=\"hljs-number\">0.6<\/span><span class=\"hljs-number\">.5<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">pip install pdf2image==<span class=\"hljs-number\">1.16<\/span><span class=\"hljs-number\">.3<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">pip install pytesseract==<span class=\"hljs-number\">0.3<\/span><span class=\"hljs-number\">.10<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\">pip install pypdf==<span class=\"hljs-number\">3.8<\/span><span class=\"hljs-number\">.1<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">pip install faiss-cpu==<span class=\"hljs-number\">1.7<\/span><span class=\"hljs-number\">.4<\/span><\/span>\n<\/code><\/pre>\n<p>\u5bf9\u4e8eLangChain\uff0c\u4f60\u53ef\u4ee5\u770b\u5230\u6211\u4eec\u6307\u5b9a\u4e86\u7248\u672c\u53f7\u3002\u8fd9\u4e2a\u5e93\u6700\u8fd1\u6b63\u5728\u63a5\u53d7\u5f88\u591a\u66f4\u65b0\uff0c\u6240\u4ee5\u4e3a\u4e86\u786e\u4fdd\u6211\u4eec\u7684\u8bbe\u7f6e\u660e\u5929\u4e5f\u80fd\u6b63\u5e38\u5de5\u4f5c\uff0c\u6700\u597d\u6307\u5b9a\u4e00\u4e2a\u6211\u4eec\u77e5\u9053\u80fd\u6b63\u5e38\u5de5\u4f5c\u7684\u7248\u672c\u3002Unstructured\u662fpdf\u52a0\u8f7d\u5668\u3001pytesseract\u548cpdf2image\u7684\u5fc5\u9700\u4f9d\u8d56\u5e93\u3002<\/p>\n<p>\u6ce8\u610f\uff1aGitHub\u5b58\u50a8\u5e93\u4e2d\u6709\u4e00\u4e2arequirements.txt\u6587\u4ef6\uff0c\u5176\u4e2d\u5305\u542b\u4e0e\u6b64\u9879\u76ee\u76f8\u5173\u7684\u6240\u6709\u7248\u672c\u3002\u4f60\u53ef\u4ee5\u5728\u5c06\u5176\u4e0b\u8f7d\u5230\u4e3b\u9879\u76ee\u6587\u4ef6\u76ee\u5f55\u540e\uff0c\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u4e00\u6b21\u6027\u8fdb\u884c\u5b89\u88c5\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">pip install -r requirements.txt<\/span>\n<\/code><\/pre>\n<p>\u8bf7\u8bb0\u4f4f\uff0c\u67d0\u4e9b\u5e93\u6709\u6839\u636e\u4f60\u5728\u865a\u62df\u73af\u5883\u4e2d\u8fd0\u884c\u7684Python\u7248\u672c\u63d0\u4f9b\u7684\u4e0d\u540c\u7248\u672c\u53ef\u7528\u3002<\/p>\n<p><strong>\u4e0b\u8f7d\u6a21\u578b<\/strong><\/p>\n<p>\u8fd9\u662f\u4e00\u4e2a\u975e\u5e38\u91cd\u8981\u7684\u6b65\u9aa4\u3002<\/p>\n<p>\u5bf9\u4e8e\u8fd9\u4e2a\u9879\u76ee\uff0c\u6211\u4eec\u5f53\u7136\u9700\u8981<\/p>\n<p>GPT4All\u6a21\u578b\u3002\u5728Nomic AI\u7f51\u7ad9\u4e0a\u63cf\u8ff0\u7684\u8fc7\u7a0b\u975e\u5e38\u590d\u6742\uff0c\u5e76\u4e14\u9700\u8981\u4e00\u4e9b\u6211\u4eec\u4e0d\u4e00\u5b9a\u90fd\u62e5\u6709\u7684\u786c\u4ef6\uff08\u6bd4\u5982\u6211\uff09\u3002\u6240\u4ee5\u8fd9\u91cc\u662f\u5df2\u7ecf\u8f6c\u6362\u5e76\u51c6\u5907\u597d\u4f7f\u7528\u7684\u6a21\u578b\u94fe\u63a5\u3002\u53ea\u9700\u70b9\u51fb\u4e0b\u8f7d\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/03b657a1db75e73073377fa0353b5ad1.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u6b63\u5982\u7b80\u4ecb\u4e2d\u7b80\u8981\u63cf\u8ff0\u7684\uff0c\u6211\u4eec\u8fd8\u9700\u8981\u5d4c\u5165\u6a21\u578b\uff0c\u8fd9\u662f\u4e00\u4e2a\u53ef\u4ee5\u5728\u6211\u4eec\u7684CPU\u4e0a\u8fd0\u884c\u800c\u4e0d\u4f1a\u5d29\u6e83\u7684\u6a21\u578b\u3002\u70b9\u51fb\u8fd9\u91cc\u7684\u94fe\u63a5\u3010<a href=\"https:\/\/link.juejin.cn?target=https%3A%2F%2Fhuggingface.co%2FPi3141%2Falpaca-native-7B-ggml%2Fblob%2F397e872bf4c83f4c642317a5bf65ce84a105786e%2Fggml-model-q4_0.bin%25E3%2580%2591%25E4%25B8%258B%25E8%25BD%25BD%25E5%25B7%25B2%25E7%25BB%258F%25E8%25BD%25AC%25E6%258D%25A2%25E4%25B8%25BA4%25E4%25BD%258D%25E5%25B9%25B6%25E5%2587%2586%25E5%25A4%2587%25E5%25A5%25BD%25E7%2594%25A8%25E4%25BD%259C%25E5%25B5%258C%25E5%2585%25A5%25E6%25A8%25A1%25E5%259E%258B%25E7%259A%2584alpaca-native-7B-ggml%25E6%25A8%25A1%25E5%259E%258B%25E3%2580%2582\" target=\"_blank\" title=\"https:\/\/huggingface.co\/Pi3141\/alpaca-native-7B-ggml\/blob\/397e872bf4c83f4c642317a5bf65ce84a105786e\/ggml-model-q4_0.bin%E3%80%91%E4%B8%8B%E8%BD%BD%E5%B7%B2%E7%BB%8F%E8%BD%AC%E6%8D%A2%E4%B8%BA4%E4%BD%8D%E5%B9%B6%E5%87%86%E5%A4%87%E5%A5%BD%E7%94%A8%E4%BD%9C%E5%B5%8C%E5%85%A5%E6%A8%A1%E5%9E%8B%E7%9A%84alpaca-native-7B-ggml%E6%A8%A1%E5%9E%8B%E3%80%82\" ref=\"nofollow noopener noreferrer\" rel=\"noopener\">huggingface.co\/Pi3141\/alpa\u2026<\/a><\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/5cb9e26849541fb5c167018cafc4cf35.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u4e3a\u4ec0\u4e48\u6211\u4eec\u9700\u8981\u5d4c\u5165\uff1f\u5982\u679c\u4f60\u8fd8\u8bb0\u5f97\u6d41\u7a0b\u56fe\u4e2d\u7684\u7b2c\u4e00\u6b65\uff0c\u5728\u6536\u96c6\u77e5\u8bc6\u5e93\u6587\u6863\u4e4b\u540e\uff0c\u6211\u4eec\u9700\u8981\u5c06\u5b83\u4eec\u5d4c\u5165\u3002LLamaCPP\u5d4c\u5165\u6765\u81eaAlpaca\u6a21\u578b\uff0c\u975e\u5e38\u9002\u5408\u8fd9\u9879\u5de5\u4f5c\uff0c\u800c\u4e14\u8fd9\u4e2a\u6a21\u578b\u4e5f\u5f88\u5c0f\uff084 GB\uff09\u3002\u987a\u4fbf\u8bf4\u4e00\u4e0b\uff0c\u4f60\u4e5f\u53ef\u4ee5\u5728\u95ee\u7b54\u73af\u8282\u4e2d\u4f7f\u7528Alpaca\u6a21\u578b\uff01<\/p>\n<p>2023.05.25\u66f4\u65b0\uff1aMani Windows\u7528\u6237\u5728\u4f7f\u7528llamaCPP\u5d4c\u5165\u65f6\u9047\u5230\u4e86\u95ee\u9898\u3002\u8fd9\u4e3b\u8981\u662f\u56e0\u4e3a\u5728\u5b89\u88c5python\u5305llama-cpp-python\u65f6\u4f7f\u7528\u4e86\u4ee5\u4e0b\u547d\u4ee4\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">pip install llama-cpp-python<\/span>\n<\/code><\/pre>\n<p>\u8fd9\u4e2a\u547d\u4ee4\u4f1a\u4ece\u6e90\u4ee3\u7801\u7f16\u8bd1\u5e93\u3002\u901a\u5e38Windows\u4e0a\u7684\u673a\u5668\u9ed8\u8ba4\u6ca1\u6709\u5b89\u88c5CMake\u6216C\u7f16\u8bd1\u5668\u3002\u4f46\u4e0d\u8981\u62c5\u5fc3\uff0c\u6709\u89e3\u51b3\u529e\u6cd5\u3002<\/p>\n<p>\u5728Windows\u4e0a\u8fd0\u884cllama-cpp-python\u7684\u5b89\u88c5\u8fc7\u7a0b\u65f6\uff0c\u9700\u8981\u7f16\u8bd1\u6e90\u4ee3\u7801\uff0c\u4f46\u7531\u4e8eWindows\u9ed8\u8ba4\u6ca1\u6709\u5b89\u88c5CMake\u548cC\u7f16\u8bd1\u5668\uff0c\u56e0\u6b64\u65e0\u6cd5\u4ece\u6e90\u4ee3\u7801\u6784\u5efa\u3002<\/p>\n<p>\u5728Mac\u7528\u6237\u4f7f\u7528Xtools\u548cLinux\u7528\u6237\u4e0a\uff0c\u901a\u5e38\u64cd\u4f5c\u7cfb\u7edf\u4e2d\u5df2\u7ecf\u5b89\u88c5\u4e86C\u7f16\u8bd1\u5668\u3002<\/p>\n<p><strong>\u4e3a\u4e86\u907f\u514d\u95ee\u9898\uff0c\u4f60\u5fc5\u987b\u4f7f\u7528\u9884\u7f16\u8bd1\u7684wheel\u3002<\/strong><\/p>\n<p>\u8bbf\u95ee<a href=\"https:\/\/link.juejin.cn?target=https%3A%2F%2Fgithub.com%2Fabetlen%2Fllama-cpp-python%2Freleases\" target=\"_blank\" title=\"https:\/\/github.com\/abetlen\/llama-cpp-python\/releases\" ref=\"nofollow noopener noreferrer\" rel=\"noopener\">github.com\/abetlen\/lla\u2026<\/a><\/p>\n<p>\u627e\u5230\u9002\u7528\u4e8e\u4f60\u67b6\u6784\u548cPython\u7248\u672c\u7684\u9884\u7f16\u8bd1\u7684wheel\u7248\u672c\uff0c\u4f60\u5fc5\u987b\u9009\u62e90.1.49\u7248\u672c\uff0c\u56e0\u4e3a\u66f4\u9ad8\u7684\u7248\u672c\u4e0d\u517c\u5bb9\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/765f68508fe79cd4461951717ad18596.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u5728\u6211\u8fd9\u91cc\uff0c\u6211\u4f7f\u7528\u7684\u662fWindows 10 64\u4f4d\uff0cPython 3.10<\/p>\n<p>\u6240\u4ee5\u6211\u7684\u6587\u4ef6\u662fllama_cpp_python-0.1.49-cp310-cp310-win_amd64.whl<\/p>\n<p>\u8fd9\u4e2a\u95ee\u9898\u5728GitHub\u5b58\u50a8\u5e93\u4e2d\u6709\u8bb0\u5f55\u3010<a href=\"https:\/\/link.juejin.cn?target=https%3A%2F%2Fgithub.com%2Ffabiomatricardi%2FGPT4All_Medium%2Fissues%2F2%25E3%2580%2591%25E3%2580%2582\" target=\"_blank\" title=\"https:\/\/github.com\/fabiomatricardi\/GPT4All_Medium\/issues\/2%E3%80%91%E3%80%82\" ref=\"nofollow noopener noreferrer\" rel=\"noopener\">github.com\/fabiomatric\u2026<\/a><\/p>\n<p>\u4e0b\u8f7d\u5b8c\u6210\u540e\uff0c\u4f60\u9700\u8981\u5c06\u8fd9\u4e24\u4e2a\u6a21\u578b\u6587\u4ef6\u653e\u5230models\u76ee\u5f55\u4e2d\uff0c\u5982\u4e0b\u6240\u793a\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/a45ad15c1d8765723d38c3d8703cb975.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u76ee\u5f55\u7ed3\u6784\u548c\u653e\u7f6e\u6a21\u578b\u6587\u4ef6\u7684\u4f4d\u7f6e<\/p>\n<h1 data-id=\"heading-3\">\u4e0eGPT4All\u7684\u57fa\u672c\u4ea4\u4e92<\/h1>\n<p>\u7531\u4e8e\u6211\u4eec\u60f3\u8981\u5bf9GPT\u6a21\u578b\u7684\u4ea4\u4e92\u8fdb\u884c\u63a7\u5236\uff0c\u6211\u4eec\u9700\u8981\u521b\u5efa\u4e00\u4e2aPython\u6587\u4ef6\uff08\u6211\u4eec\u79f0\u4e4b\u4e3apygpt4all_test.py\uff09\uff0c\u5bfc\u5165\u4f9d\u8d56\u9879\u5e76\u7ed9\u6a21\u578b\u53d1\u9001\u6307\u4ee4\u3002\u4f60\u4f1a\u53d1\u73b0\u8fd9\u5f88\u5bb9\u6613\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-keyword\">from<\/span> pygpt4all.models.gpt4all <span class=\"hljs-keyword\">import<\/span> GPT4All<\/span>\n<\/code><\/pre>\n<p>\u8fd9\u662f\u6211\u4eec\u6a21\u578b\u7684Python\u7ed1\u5b9a\u3002\u73b0\u5728\u6211\u4eec\u53ef\u4ee5\u8c03\u7528\u5b83\u5e76\u5f00\u59cb\u63d0\u95ee\u3002\u8ba9\u6211\u4eec\u8bd5\u4e00\u4e2a\u6709\u521b\u610f\u7684\u95ee\u9898\u3002<\/p>\n<p>\u6211\u4eec\u521b\u5efa\u4e00\u4e2a\u51fd\u6570\u6765\u8bfb\u53d6\u6a21\u578b\u7684\u56de\u8c03\uff0c\u5e76\u8981\u6c42GPT4All\u5b8c\u6210\u6211\u4eec\u7684\u53e5\u5b50\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">new_text_callback<\/span>(<span class=\"hljs-params\">text<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">    <span class=\"hljs-built_in\">print<\/span>(text, end=<span class=\"hljs-string\">\"\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\">model = GPT4All(<span class=\"hljs-string\">'.\/models\/gpt4all-converted.bin'<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">model.generate(<span class=\"hljs-string\">\"Once upon a time, \"<\/span>, n_predict=<span class=\"hljs-number\">55<\/span>, new_text_callback=new_text_callback)<\/span>\n<\/code><\/pre>\n<p>\u7b2c\u4e00\u6761\u8bed\u53e5\u544a\u8bc9\u6211\u4eec\u7684\u7a0b\u5e8f\u5728\u54ea\u91cc\u627e\u5230\u6a21\u578b\uff08\u8bb0\u4f4f\u6211\u4eec\u5728\u4e0a\u9762\u7684\u90e8\u5206\u6240\u505a\u7684\u4e8b\u60c5\uff09\u3002<\/p>\n<p>\u7b2c\u4e8c\u6761\u8bed\u53e5\u8981\u6c42\u6a21\u578b\u751f\u6210\u4e00\u4e2a\u56de\u7b54\uff0c\u5e76\u5b8c\u6210\u6211\u4eec\u7684\u63d0\u793a\u6587\u672c &#8220;Once upon a time,&#8221;\u3002<\/p>\n<p>\u8981\u8fd0\u884c\u5b83\uff0c\u8bf7\u786e\u4fdd\u865a\u62df\u73af\u5883\u4ecd\u5904\u4e8e\u6fc0\u6d3b\u72b6\u6001\uff0c\u7136\u540e\u7b80\u5355\u5730\u8fd0\u884c\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">python3 pygpt4all_test.py<\/span>\n<\/code><\/pre>\n<p>\u4f60\u5e94\u8be5\u4f1a\u770b\u5230\u6a21\u578b\u6b63\u5728\u52a0\u8f7d\u7684\u6587\u672c\u548c\u53e5\u5b50\u7684\u5b8c\u6210\u3002\u6839\u636e\u4f60\u7684\u786c\u4ef6\u8d44\u6e90\uff0c\u8fd9\u53ef\u80fd\u9700\u8981\u4e00\u4e9b\u65f6\u95f4\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/966b927acc4e6dd5f9650a9c27140f45.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u7ed3\u679c\u53ef\u80fd\u4e0e\u4f60\u7684\u4e0d\u540c&#8230; \u4f46\u5bf9\u6211\u4eec\u6765\u8bf4\u91cd\u8981\u7684\u662f\u5b83\u6b63\u5728\u5de5\u4f5c\uff0c\u6211\u4eec\u53ef\u4ee5\u7ee7\u7eed\u4f7f\u7528LangChain\u521b\u5efa\u4e00\u4e9b\u9ad8\u7ea7\u529f\u80fd\u3002<\/p>\n<p>\u6ce8\u610f\uff08\u66f4\u65b0\u4e8e2023.05.23\uff09\uff1a\u5982\u679c\u4f60\u9047\u5230\u4e0epygpt4all\u76f8\u5173\u7684\u9519\u8bef\uff0c\u8bf7\u67e5\u770b\u6545\u969c\u6392\u9664\u90e8\u5206\uff0c\u5176\u4e2d\u63d0\u4f9b\u4e86\u7531Rajneesh Aggarwal\u6216Oscar Jeong\u7ed9\u51fa\u7684\u89e3\u51b3\u65b9\u6848\u3002<\/p>\n<h1 data-id=\"heading-4\">LangChain\u5728GPT4All\u4e0a\u7684\u6a21\u677f<\/h1>\n<p>LangChain\u6846\u67b6\u662f\u4e00\u4e2a\u975e\u5e38\u5f3a\u5927\u7684\u5e93\u3002\u5b83\u63d0\u4f9b\u4e86\u7ec4\u4ef6\uff0c\u4ee5\u4e00\u79cd\u6613\u4e8e\u4f7f\u7528\u7684\u65b9\u5f0f\u4e0e\u8bed\u8a00\u6a21\u578b\u8fdb\u884c\u4ea4\u4e92\uff0c\u5e76\u4e14\u8fd8\u63d0\u4f9b\u4e86Chains\u3002\u53ef\u4ee5\u5c06Chains\u89c6\u4e3a\u4ee5\u7279\u5b9a\u65b9\u5f0f\u7ec4\u88c5\u8fd9\u4e9b\u7ec4\u4ef6\uff0c\u4ee5\u6700\u597d\u5730\u5b8c\u6210\u7279\u5b9a\u7684\u7528\u4f8b\u3002\u5b83\u4eec\u65e8\u5728\u6210\u4e3a\u4e00\u4e2a\u66f4\u9ad8\u7ea7\u7684\u63a5\u53e3\uff0c\u4f7f\u4eba\u4eec\u80fd\u591f\u8f7b\u677e\u5730\u5f00\u59cb\u4f7f\u7528\u7279\u5b9a\u7684\u7528\u4f8b\u3002\u8fd9\u4e9bChains\u4e5f\u53ef\u4ee5\u8fdb\u884c\u5b9a\u5236\u3002<\/p>\n<p>\u5728\u6211\u4eec\u7684\u4e0b\u4e00\u4e2aPython\u6d4b\u8bd5\u4e2d\uff0c\u6211\u4eec\u5c06\u4f7f\u7528\u4e00\u4e2aPrompt Template\u3002\u8bed\u8a00\u6a21\u578b\u63a5\u53d7\u6587\u672c\u4f5c\u4e3a\u8f93\u5165\uff0c\u8fd9\u6bb5\u6587\u672c\u901a\u5e38\u88ab\u79f0\u4e3aprompt\u3002\u901a\u5e38\u60c5\u51b5\u4e0b\uff0c\u8fd9\u4e0d\u4ec5\u4ec5\u662f\u4e00\u4e2a\u786c\u7f16\u7801\u7684\u5b57\u7b26\u4e32\uff0c\u800c\u662f\u6a21\u677f\u3001\u793a\u4f8b\u548c\u7528\u6237\u8f93\u5165\u7684\u7ec4\u5408\u3002LangChain\u63d0\u4f9b\u4e86\u591a\u4e2a\u7c7b\u548c\u51fd\u6570\uff0c\u4f7f\u6784\u5efa\u548c\u5904\u7406prompt\u53d8\u5f97\u7b80\u5355\u3002\u8ba9\u6211\u4eec\u770b\u770b\u5982\u4f55\u5b9e\u73b0\u3002<\/p>\n<p>\u521b\u5efa\u4e00\u4e2a\u65b0\u7684Python\u6587\u4ef6\uff0c\u547d\u540d\u4e3amy_langchain.py\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># Import of langchain Prompt Template and Chain<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><span class=\"hljs-keyword\">from<\/span> langchain <span class=\"hljs-keyword\">import<\/span> PromptTemplate, LLMChain<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-comment\"># Import llm to be able to interact with GPT4All directly from langchain<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\"><span class=\"hljs-keyword\">from<\/span> langchain.llms <span class=\"hljs-keyword\">import<\/span> GPT4All<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-comment\"># Callbacks manager is required for the response handling <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.base <span class=\"hljs-keyword\">import<\/span> CallbackManager<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.streaming_stdout <span class=\"hljs-keyword\">import<\/span> StreamingStdOutCallbackHandler<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\">local_path = <span class=\"hljs-string\">'.\/models\/gpt4all-converted.bin'<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\">callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])<\/span>\n<\/code><\/pre>\n<p>\u6211\u4eec\u4eceLangChain\u4e2d\u5bfc\u5165\u4e86Prompt Template\u548cChain\uff0c\u4ee5\u53caGPT4All llm\u7c7b\uff0c\u4ee5\u4fbf\u80fd\u591f\u76f4\u63a5\u4e0e\u6211\u4eec\u7684GPT\u6a21\u578b\u8fdb\u884c\u4ea4\u4e92\u3002<\/p>\n<p>\u7136\u540e\uff0c\u5728\u8bbe\u7f6e\u4e86llm\u8def\u5f84\u4e4b\u540e\uff08\u4e0e\u4e4b\u524d\u4e00\u6837\uff09\uff0c\u6211\u4eec\u5b9e\u4f8b\u5316\u4e86\u56de\u8c03\u7ba1\u7406\u5668\uff0c\u4ee5\u4fbf\u80fd\u591f\u6355\u83b7\u6211\u4eec\u67e5\u8be2\u7684\u54cd\u5e94\u3002<\/p>\n<p>\u521b\u5efa\u4e00\u4e2a\u6a21\u677f\u975e\u5e38\u7b80\u5355\uff1a\u6839\u636e\u6587\u6863\u6559\u7a0b\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u5982\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">template = <span class=\"hljs-string\">\"\"\"Question: {question}<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">Answer: Let's think step by step on it.<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">\"\"\"<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">prompt = PromptTemplate(template=template, input_variables=[<span class=\"hljs-string\">\"question\"<\/span>])<\/span>\n<\/code><\/pre>\n<p>template\u53d8\u91cf\u662f\u4e00\u4e2a\u591a\u884c\u5b57\u7b26\u4e32\uff0c\u5305\u542b\u4e86\u4e0e\u6a21\u578b\u7684\u4ea4\u4e92\u7ed3\u6784\uff1a\u5728\u82b1\u62ec\u53f7\u4e2d\u63d2\u5165\u6a21\u677f\u7684\u5916\u90e8\u53d8\u91cf\uff0c\u5bf9\u4e8e\u6211\u4eec\u7684\u60c5\u51b5\u5c31\u662f\u95ee\u9898\u3002<\/p>\n<p>\u7531\u4e8e\u5b83\u662f\u4e00\u4e2a\u53d8\u91cf\uff0c\u4f60\u53ef\u4ee5\u51b3\u5b9a\u5b83\u662f\u4e00\u4e2a\u786c\u7f16\u7801\u7684\u95ee\u9898\u8fd8\u662f\u7528\u6237\u8f93\u5165\u7684\u95ee\u9898\uff1a\u8fd9\u91cc\u662f\u4e24\u4e2a\u793a\u4f8b\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># Hardcoded question<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">question = <span class=\"hljs-string\">\"What Formula 1 pilot won the championship in the year Leonardo di Caprio was born?\"<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-comment\"># User input question...<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">question = <span class=\"hljs-built_in\">input<\/span>(<span class=\"hljs-string\">\"Enter your question: \"<\/span>)<\/span>\n<\/code><\/pre>\n<p>\u5bf9\u4e8e\u6211\u4eec\u7684\u6d4b\u8bd5\u8fd0\u884c\uff0c\u6211\u4eec\u5c06\u6ce8\u91ca\u6389\u7528\u6237\u8f93\u5165\u7684\u95ee\u9898\u3002\u73b0\u5728\u6211\u4eec\u53ea\u9700\u8981\u5c06\u6211\u4eec\u7684\u6a21\u677f\u3001\u95ee\u9898\u548c\u8bed\u8a00\u6a21\u578b\u8fde\u63a5\u5728\u4e00\u8d77\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">template = <span class=\"hljs-string\">\"\"\"Question: {question}<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">Answer: Let's think step by step on it.<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">\"\"\"<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">prompt = PromptTemplate(template=template, input_variables=[<span class=\"hljs-string\">\"question\"<\/span>])<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-comment\"># initialize the GPT4All instance<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">llm = GPT4All(model=local_path, callback_manager=callback_manager, verbose=<span class=\"hljs-literal\">True<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\"><span class=\"hljs-comment\"># link the language model with our prompt template<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\">llm_chain = LLMChain(prompt=prompt, llm=llm)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\"><span class=\"hljs-comment\"># Hardcoded question<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\">question = <span class=\"hljs-string\">\"What Formula 1 pilot won the championship in the year Leonardo di Caprio was born?\"<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\"><span class=\"hljs-comment\"># User imput question...<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\"><span class=\"hljs-comment\"># question = input(\"Enter your question: \")<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\"><span class=\"hljs-comment\">#Run the query and get the results<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\">llm_chain.run(question)<\/span>\n<\/code><\/pre>\n<p>\u8bb0\u5f97\u9a8c\u8bc1\u4f60\u7684\u865a\u62df\u73af\u5883\u4ecd\u7136\u5904\u4e8e\u6fc0\u6d3b\u72b6\u6001\uff0c\u5e76\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">python3 my_langchain.py<\/span>\n<\/code><\/pre>\n<p>\u4f60\u53ef\u80fd\u4f1a\u5f97\u5230\u4e0e\u6211\u4e0d\u540c\u7684\u7ed3\u679c\u3002\u4ee4\u4eba\u60ca\u5947\u7684\u662f\uff0c\u4f60\u53ef\u4ee5\u770b\u5230GPT4All\u5728\u5c1d\u8bd5\u4e3a\u4f60\u627e\u5230\u7b54\u6848\u65f6\u6240\u9075\u5faa\u7684\u6574\u4e2a\u63a8\u7406\u8fc7\u7a0b\u3002\u8c03\u6574\u95ee\u9898\u53ef\u80fd\u4f1a\u5f97\u5230\u66f4\u597d\u7684\u7ed3\u679c\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/0c8791ec26b04beec136207a51a94fe5.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<h1 data-id=\"heading-5\">\u4f7f\u7528LangChain\u548cGPT4All\u56de\u7b54\u5173\u4e8e\u6587\u4ef6\u7684\u95ee\u9898<\/h1>\n<p>\u5728\u8fd9\u91cc\uff0c\u6211\u4eec\u5f00\u59cb\u4e86\u4ee4\u4eba\u60ca\u5947\u7684\u90e8\u5206\uff0c\u56e0\u4e3a\u6211\u4eec\u5c06\u4f7f\u7528GPT4All\u4f5c\u4e3a\u4e00\u4e2a\u804a\u5929\u673a\u5668\u4eba\u6765\u56de\u7b54\u6211\u4eec\u7684\u95ee\u9898\u3002<\/p>\n<p>\u6839\u636e\u4e0eGPT4All\u8fdb\u884c\u95ee\u7b54\u7684\u5de5\u4f5c\u6d41\u7a0b\uff0c\u6211\u4eec\u9700\u8981\u52a0\u8f7d\u6211\u4eec\u7684PDF\u6587\u4ef6\uff0c\u5e76\u5c06\u5176\u5206\u6210\u5757\u3002\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u9700\u8981\u4e3a\u6211\u4eec\u7684\u5d4c\u5165\u5411\u91cf\u51c6\u5907\u4e00\u4e2a\u5411\u91cf\u5b58\u50a8\u5e93\u3002\u6211\u4eec\u9700\u8981\u5c06\u6211\u4eec\u7684\u5206\u5757\u6587\u6863\u8f93\u5165\u5230\u5411\u91cf\u5b58\u50a8\u5e93\u4e2d\u8fdb\u884c\u4fe1\u606f\u68c0\u7d22\uff0c\u7136\u540e\u5c06\u5b83\u4eec\u4e0e\u8be5\u6570\u636e\u5e93\u4e0a\u7684\u76f8\u4f3c\u6027\u641c\u7d22\u4e00\u8d77\u4f5c\u4e3aLLM\u67e5\u8be2\u7684\u4e0a\u4e0b\u6587\u8fdb\u884c\u5d4c\u5165\u3002<\/p>\n<p>\u4e3a\u6b64\uff0c\u6211\u4eec\u5c06\u76f4\u63a5\u4f7f\u7528Langchain\u5e93\u4e2d\u7684FAISS\u3002FAISS\u662fFacebook AI Research\u5f00\u53d1\u7684\u5f00\u6e90\u5e93\uff0c\u65e8\u5728\u5feb\u901f\u5728\u5927\u89c4\u6a21\u9ad8\u7ef4\u6570\u636e\u96c6\u4e2d\u67e5\u627e\u76f8\u4f3c\u9879\u3002\u5b83\u63d0\u4f9b\u7d22\u5f15\u548c\u641c\u7d22\u65b9\u6cd5\uff0c\u4f7f\u5f97\u5728\u6570\u636e\u96c6\u4e2d\u5feb\u901f\u627e\u5230\u6700\u76f8\u4f3c\u7684\u9879\u53d8\u5f97\u66f4\u52a0\u7b80\u5355\u548c\u5feb\u901f\u3002\u5bf9\u6211\u4eec\u6765\u8bf4\u7279\u522b\u65b9\u4fbf\u7684\u662f\uff0c\u5b83\u7b80\u5316\u4e86\u4fe1\u606f\u68c0\u7d22\uff0c\u5e76\u5141\u8bb8\u6211\u4eec\u672c\u5730\u4fdd\u5b58\u521b\u5efa\u7684\u6570\u636e\u5e93\uff1a\u8fd9\u610f\u5473\u7740\u5728\u7b2c\u4e00\u6b21\u521b\u5efa\u540e\uff0c\u5c06\u975e\u5e38\u5feb\u901f\u5730\u52a0\u8f7d\u6570\u636e\u5e93\u4ee5\u4f9b\u4ee5\u540e\u4f7f\u7528\u3002<\/p>\n<p><strong>\u521b\u5efa\u5411\u91cf\u7d22\u5f15\u6570\u636e\u5e93<\/strong><\/p>\n<p>\u521b\u5efa\u4e00\u4e2a\u65b0\u7684\u6587\u4ef6\u53eb\u505a&nbsp;my_knowledge_qna.py\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-keyword\">from<\/span> langchain <span class=\"hljs-keyword\">import<\/span> PromptTemplate, LLMChain<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><span class=\"hljs-keyword\">from<\/span> langchain.llms <span class=\"hljs-keyword\">import<\/span> GPT4All<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.base <span class=\"hljs-keyword\">import<\/span> CallbackManager<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.streaming_stdout <span class=\"hljs-keyword\">import<\/span> StreamingStdOutCallbackHandler<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\"><span class=\"hljs-comment\"># function for loading only TXT files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> TextLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-comment\"># text splitter for create chunks<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\"><span class=\"hljs-keyword\">from<\/span> langchain.text_splitter <span class=\"hljs-keyword\">import<\/span> RecursiveCharacterTextSplitter<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\"><span class=\"hljs-comment\"># to be able to load the pdf files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> UnstructuredPDFLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> PyPDFLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> DirectoryLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\"><span class=\"hljs-comment\"># Vector Store Index to create our database about our knowledge<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\"><span class=\"hljs-keyword\">from<\/span> langchain.indexes <span class=\"hljs-keyword\">import<\/span> VectorstoreIndexCreator<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\"><span class=\"hljs-comment\"># LLamaCpp embeddings from the Alpaca model<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\"><span class=\"hljs-keyword\">from<\/span> langchain.embeddings <span class=\"hljs-keyword\">import<\/span> LlamaCppEmbeddings<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"22\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"23\"><span class=\"hljs-comment\"># FAISS  library for similaarity search<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"24\"><span class=\"hljs-keyword\">from<\/span> langchain.vectorstores.faiss <span class=\"hljs-keyword\">import<\/span> FAISS<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"25\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"26\"><span class=\"hljs-keyword\">import<\/span> os  <span class=\"hljs-comment\">#for interaaction with the files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"27\"><span class=\"hljs-keyword\">import<\/span> datetime<\/span>\n<\/code><\/pre>\n<p>\u7b2c\u4e00\u7ec4\u5e93\u4e0e\u4e4b\u524d\u4f7f\u7528\u7684\u76f8\u540c\uff1a\u53e6\u5916\u6211\u4eec\u4f7f\u7528Langchain\u8fdb\u884c\u5411\u91cf\u5b58\u50a8\u7d22\u5f15\u7684\u521b\u5efa\uff0c\u4f7f\u7528LlamaCppEmbeddings\u4e0e\u6211\u4eec\u7684Alpaca\u6a21\u578b\u8fdb\u884c\u4ea4\u4e92\uff08\u4f7f\u7528cpp\u5e93\u8fdb\u884c4\u4f4d\u91cf\u5316\u548c\u7f16\u8bd1\uff09\uff0c\u4ee5\u53caPDF\u52a0\u8f7d\u5668\u3002<\/p>\n<p>\u6211\u4eec\u8fd8\u8981\u52a0\u8f7d\u5177\u6709\u81ea\u5df1\u8def\u5f84\u7684LLMs\uff1a\u4e00\u4e2a\u7528\u4e8e\u5d4c\u5165\u548c\u4e00\u4e2a\u7528\u4e8e\u6587\u672c\u751f\u6210\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># assign the path for the 2 models GPT4All and Alpaca for the embeddings <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">gpt4all_path = <span class=\"hljs-string\">'.\/models\/gpt4all-converted.bin'<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">llama_path = <span class=\"hljs-string\">'.\/models\/ggml-model-q4_0.bin'<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-comment\"># Calback manager for handling the calls with  the model<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-comment\"># create the embedding object<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">embeddings = LlamaCppEmbeddings(model_path=llama_path)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-comment\"># create the GPT4All llm object<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\">llm = GPT4All(model=gpt4all_path, callback_manager=callback_manager, verbose=<span class=\"hljs-literal\">True<\/span>)<\/span>\n<\/code><\/pre>\n<p>\u4e3a\u4e86\u6d4b\u8bd5\uff0c\u8ba9\u6211\u4eec\u770b\u770b\u662f\u5426\u6210\u529f\u8bfb\u53d6\u4e86\u6240\u6709\u7684PDF\u6587\u4ef6\uff1a\u7b2c\u4e00\u6b65\u662f\u58f0\u660e\u4e09\u4e2a\u51fd\u6570\uff0c\u7528\u4e8e\u5904\u7406\u6bcf\u4e2a\u5355\u72ec\u7684\u6587\u6863\u3002\u7b2c\u4e00\u4e2a\u51fd\u6570\u662f\u5c06\u63d0\u53d6\u7684\u6587\u672c\u5206\u5272\u6210\u5757\uff0c\u7b2c\u4e8c\u4e2a\u51fd\u6570\u662f\u521b\u5efa\u5e26\u6709\u5143\u6570\u636e\u7684\u5411\u91cf\u7d22\u5f15\uff08\u4f8b\u5982\u9875\u7801\u7b49\uff09\uff0c\u6700\u540e\u4e00\u4e2a\u51fd\u6570\u662f\u7528\u4e8e\u6d4b\u8bd5\u76f8\u4f3c\u6027\u641c\u7d22\uff08\u7a0d\u540e\u6211\u5c06\u66f4\u8be6\u7ec6\u5730\u89e3\u91ca\uff09\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># Split text <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">split_chunks<\/span>(<span class=\"hljs-params\">sources<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">    chunks = []<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\">    splitter = RecursiveCharacterTextSplitter(chunk_size=<span class=\"hljs-number\">256<\/span>, chunk_overlap=<span class=\"hljs-number\">32<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">    <span class=\"hljs-keyword\">for<\/span> chunk <span class=\"hljs-keyword\">in<\/span> splitter.split_documents(sources):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">        chunks.append(chunk)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\">    <span class=\"hljs-keyword\">return<\/span> chunks<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">create_index<\/span>(<span class=\"hljs-params\">chunks<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\">    texts = [doc.page_content <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> chunks]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\">    metadatas = [doc.metadata <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> chunks]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\">    search_index = FAISS.from_texts(texts, embeddings, metadatas=metadatas)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\">    <span class=\"hljs-keyword\">return<\/span> search_index<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">similarity_search<\/span>(<span class=\"hljs-params\">query, index<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\">    <span class=\"hljs-comment\"># k is the number of similarity searched that matches the query<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\">    <span class=\"hljs-comment\"># default is 4<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"22\">    matched_docs = index.similarity_search(query, k=<span class=\"hljs-number\">3<\/span>) <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"23\">    sources = []<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"24\">    <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> matched_docs:<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"25\">        sources.append(<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"26\">            {<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"27\">                <span class=\"hljs-string\">\"page_content\"<\/span>: doc.page_content,<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"28\">                <span class=\"hljs-string\">\"metadata\"<\/span>: doc.metadata,<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"29\">            }<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"30\">        )<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"31\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"32\">    <span class=\"hljs-keyword\">return<\/span> matched_docs, sources<\/span>\n<\/code><\/pre>\n<p>\u73b0\u5728\u6211\u4eec\u53ef\u4ee5\u6d4b\u8bd5\u6587\u6863\u76ee\u5f55\u4e2d\u7684\u7d22\u5f15\u751f\u6210\uff1a\u6211\u4eec\u9700\u8981\u5c06\u6240\u6709\u7684PDF\u6587\u4ef6\u653e\u5728\u90a3\u91cc\u3002Langchain\u8fd8\u6709\u4e00\u79cd\u52a0\u8f7d\u6574\u4e2a\u6587\u4ef6\u5939\u7684\u65b9\u6cd5\uff0c\u4e0d\u8bba\u6587\u4ef6\u7c7b\u578b\u5982\u4f55\uff1a\u7531\u4e8e\u540e\u7eed\u5904\u7406\u6bd4\u8f83\u590d\u6742\uff0c\u6211\u5c06\u5728\u4e0b\u4e00\u7bc7\u5173\u4e8eLaMini\u6a21\u578b\u7684\u6587\u7ae0\u4e2d\u4ecb\u7ecd\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/9af4b88b65b4d7ed27668f19678b85d1.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u6211\u4eec\u5c06\u628a\u8fd9\u4e9b\u51fd\u6570\u5e94\u7528\u5230\u5217\u8868\u4e2d\u7684\u7b2c\u4e00\u4e2a\u6587\u6863\u4e0a\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># get the list of pdf files from the docs directory into a list  format<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">pdf_folder_path = <span class=\"hljs-string\">'.\/docs'<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">doc_list = [s <span class=\"hljs-keyword\">for<\/span> s <span class=\"hljs-keyword\">in<\/span> os.listdir(pdf_folder_path) <span class=\"hljs-keyword\">if<\/span> s.endswith(<span class=\"hljs-string\">'.pdf'<\/span>)]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\">num_of_docs = <span class=\"hljs-built_in\">len<\/span>(doc_list)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\"><span class=\"hljs-comment\"># create a loader for the PDFs from the path<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">loader = PyPDFLoader(os.path.join(pdf_folder_path, doc_list[<span class=\"hljs-number\">0<\/span>]))<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-comment\"># load the documents with Langchain<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">docs = loader.load()<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-comment\"># Split in chunks<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\">chunks = split_chunks(docs)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\"><span class=\"hljs-comment\"># create the db vector index<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\">db0 = create_index(chunks)<\/span>\n<\/code><\/pre>\n<p>\u5728\u7b2c\u4e00\u884c\u4e2d\uff0c\u6211\u4eec\u4f7f\u7528os\u5e93\u6765\u83b7\u53d6docs\u76ee\u5f55\u4e2d\u7684PDF\u6587\u4ef6\u5217\u8868\u3002\u7136\u540e\uff0c\u6211\u4eec\u4f7f\u7528Langchain\u52a0\u8f7ddocs\u6587\u4ef6\u5939\u4e2d\u7684\u7b2c\u4e00\u4e2a\u6587\u6863\uff08doc_list[0]\uff09\uff0c\u5c06\u5176\u5206\u5272\u6210\u5757\uff0c\u7136\u540e\u4f7f\u7528LLama\u5d4c\u5165\u521b\u5efa\u5411\u91cf\u6570\u636e\u5e93\u3002<\/p>\n<p>\u6b63\u5982\u60a8\u6240\u770b\u5230\u7684\uff0c\u6211\u4eec\u4f7f\u7528\u4e86pyPDF\u65b9\u6cd5\u3002\u8fd9\u4e2a\u65b9\u6cd5\u7a0d\u5fae\u590d\u6742\u4e00\u4e9b\uff0c\u56e0\u4e3a\u60a8\u9700\u8981\u9010\u4e2a\u52a0\u8f7d\u6587\u4ef6\uff0c\u4f46\u662f\u4f7f\u7528pypdf\u5c06PDF\u52a0\u8f7d\u5230\u6587\u6863\u6570\u7ec4\u4e2d\u4f7f\u60a8\u80fd\u591f\u5f97\u5230\u4e00\u4e2a\u6570\u7ec4\uff0c\u5176\u4e2d\u6bcf\u4e2a\u6587\u6863\u90fd\u5305\u542b\u9875\u9762\u5185\u5bb9\u548c\u5e26\u6709\u9875\u7801\u7684\u5143\u6570\u636e\u3002\u5f53\u60a8\u60f3\u8981\u77e5\u9053\u6211\u4eec\u5c06\u901a\u8fc7\u67e5\u8be2\u63d0\u4f9b\u7ed9GPT4All\u7684\u4e0a\u4e0b\u6587\u7684\u6765\u6e90\u65f6\uff0c\u8fd9\u975e\u5e38\u65b9\u4fbf\u3002\u4ee5\u4e0b\u662f\u6765\u81eareadthedocs\u7684\u793a\u4f8b\uff1a<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/6d425b260ebcc1359f0e382d6b5767d4.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u7136\u540e\u6267\u884c\u4e0b\u9762\u7684\u547d\u4ee4\u8fd0\u884c\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">python3 my_knowledge_qna.py<\/span>\n<\/code><\/pre>\n<p>\u5728\u52a0\u8f7d\u7528\u4e8e\u5d4c\u5165\u7684\u6a21\u578b\u4e4b\u540e\uff0c\u60a8\u5c06\u770b\u5230\u4ee4\u724c\u5728\u7d22\u5f15\u4e2d\u7684\u5de5\u4f5c\u65b9\u5f0f\uff1a\u4e0d\u8981\u60ca\u614c\uff0c\u56e0\u4e3a\u8fd9\u9700\u8981\u65f6\u95f4\uff0c\u7279\u522b\u662f\u5982\u679c\u60a8\u53ea\u5728CPU\u4e0a\u8fd0\u884c\uff0c\u5c31\u50cf\u6211\u4e00\u6837\uff08\u8fd9\u9700\u89818\u5206\u949f\uff09\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/6c3553c69aafcd5b29167bb9a3e1995a.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u6b63\u5982\u6211\u4e4b\u524d\u89e3\u91ca\u7684\uff0cpyPDF\u65b9\u6cd5\u8f83\u6162\uff0c\u4f46\u4e3a\u6211\u4eec\u63d0\u4f9b\u4e86\u7528\u4e8e\u76f8\u4f3c\u6027\u641c\u7d22\u7684\u989d\u5916\u6570\u636e\u3002\u4e3a\u4e86\u904d\u5386\u6240\u6709\u6587\u4ef6\uff0c\u6211\u4eec\u5c06\u4f7f\u7528FAISS\u4e2d\u7684\u4e00\u79cd\u4fbf\u6377\u65b9\u6cd5\uff0c\u5b83\u5141\u8bb8\u6211\u4eec\u5c06\u4e0d\u540c\u7684\u6570\u636e\u5e93\u5408\u5e76\u5728\u4e00\u8d77\u3002\u73b0\u5728\u6211\u4eec\u505a\u7684\u662f\u4f7f\u7528\u4e0a\u9762\u7684\u4ee3\u7801\u751f\u6210\u7b2c\u4e00\u4e2a\u6570\u636e\u5e93\uff08\u6211\u4eec\u5c06\u5176\u79f0\u4e3adb0\uff09\uff0c\u7136\u540e\u4f7f\u7528for\u5faa\u73af\u521b\u5efa\u5217\u8868\u4e2d\u4e0b\u4e00\u4e2a\u6587\u4ef6\u7684\u7d22\u5f15\uff0c\u5e76\u7acb\u5373\u5c06\u5176\u4e0edb0\u5408\u5e76\u3002<\/p>\n<p>\u4ee5\u4e0b\u662f\u4ee3\u7801\uff1a\u8bf7\u6ce8\u610f\uff0c\u6211\u6dfb\u52a0\u4e86\u4e00\u4e9b\u65e5\u5fd7\uff0c\u4ee5\u4fbf\u4f7f\u7528datetime.datetime.now()\u63d0\u4f9b\u8fdb\u5ea6\u72b6\u6001\uff0c\u5e76\u6253\u5370\u7ed3\u675f\u65f6\u95f4\u548c\u5f00\u59cb\u65f6\u95f4\u7684\u5dee\u503c\u6765\u8ba1\u7b97\u64cd\u4f5c\u6240\u9700\u7684\u65f6\u95f4\uff08\u5982\u679c\u60a8\u4e0d\u559c\u6b22\uff0c\u53ef\u4ee5\u5c06\u5176\u5220\u9664\uff09\u3002<\/p>\n<p>\u5408\u5e76\u6307\u4ee4\u5982\u4e0b\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># merge dbi with the existing db0db0.merge_from(dbi)<\/span><\/span>\n<\/code><\/pre>\n<p>\u6700\u540e\u4e00\u6761\u6307\u4ee4\u662f\u5c06\u6211\u4eec\u7684\u6570\u636e\u5e93\u4fdd\u5b58\u5230\u672c\u5730\uff1a\u6574\u4e2a\u751f\u6210\u8fc7\u7a0b\u53ef\u80fd\u9700\u8981\u6570\u5c0f\u65f6\uff08\u53d6\u51b3\u4e8e\u6587\u6863\u7684\u6570\u91cf\uff09\uff0c\u6240\u4ee5\u6211\u4eec\u53ea\u9700\u6267\u884c\u4e00\u6b21\u8fd9\u4e2a\u64cd\u4f5c\u975e\u5e38\u597d\uff01<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># Save the databasae locallydb0.save_local(\"my_faiss_index\")<\/span><\/span>\n<\/code><\/pre>\n<p>\u4ee5\u4e0b\u662f\u5b8c\u6574\u7684\u4ee3\u7801\u3002\u5f53\u6211\u4eec\u4e0eGPT4All\u4e92\u52a8\u5e76\u76f4\u63a5\u4ece\u6587\u4ef6\u5939\u52a0\u8f7d\u7d22\u5f15\u65f6\uff0c\u6211\u4eec\u5c06\u5bf9\u5176\u4e2d\u8bb8\u591a\u90e8\u5206\u8fdb\u884c\u6ce8\u91ca\u3002<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># get the list of pdf files from the docs directory into a list  format<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">pdf_folder_path = <span class=\"hljs-string\">'.\/docs'<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\">doc_list = [s <span class=\"hljs-keyword\">for<\/span> s <span class=\"hljs-keyword\">in<\/span> os.listdir(pdf_folder_path) <span class=\"hljs-keyword\">if<\/span> s.endswith(<span class=\"hljs-string\">'.pdf'<\/span>)]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\">num_of_docs = <span class=\"hljs-built_in\">len<\/span>(doc_list)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\"><span class=\"hljs-comment\"># create a loader for the PDFs from the path<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">general_start = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"starting the loop...\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">loop_start = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"generating fist vector database and then iterate with .merge_from\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\">loader = PyPDFLoader(os.path.join(pdf_folder_path, doc_list[<span class=\"hljs-number\">0<\/span>]))<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\">docs = loader.load()<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\">chunks = split_chunks(docs)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\">db0 = create_index(chunks)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"Main Vector database created. Start iteration and merging...\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><span class=\"hljs-keyword\">for<\/span> i <span class=\"hljs-keyword\">in<\/span> <span class=\"hljs-built_in\">range<\/span>(<span class=\"hljs-number\">1<\/span>,num_of_docs):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\">    <span class=\"hljs-built_in\">print<\/span>(doc_list[i])<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\">    <span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"loop position <span class=\"hljs-subst\">{i}<\/span>\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\">    loader = PyPDFLoader(os.path.join(pdf_folder_path, doc_list[i]))<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\">    start = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\">    docs = loader.load()<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\">    chunks = split_chunks(docs)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"22\">    dbi = create_index(chunks)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"23\">    <span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"start merging with db0...\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"24\">    db0.merge_from(dbi)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"25\">    end = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"26\">    elapsed = end - start <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"27\">    <span class=\"hljs-comment\">#total time<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"28\">    <span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"completed in <span class=\"hljs-subst\">{elapsed}<\/span>\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"29\">    <span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"-----------------------------------\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"30\">loop_end = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"31\">loop_elapsed = loop_end - loop_start <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"32\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"All documents processed in <span class=\"hljs-subst\">{loop_elapsed}<\/span>\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"33\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"the daatabase is done with <span class=\"hljs-subst\">{num_of_docs}<\/span> subset of db index\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"34\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"-----------------------------------\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"35\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"Merging completed\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"36\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"-----------------------------------\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"37\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"Saving Merged Database Locally\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"38\"><span class=\"hljs-comment\"># Save the databasae locally<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"39\">db0.save_local(<span class=\"hljs-string\">\"my_faiss_index\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"40\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"-----------------------------------\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"41\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"merged database saved as my_faiss_index\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"42\">general_end = datetime.datetime.now() <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"43\">general_elapsed = general_end - general_start <span class=\"hljs-comment\">#not used now but useful<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"44\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"All indexing completed in <span class=\"hljs-subst\">{general_elapsed}<\/span>\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"45\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"-----------------------------------\"<\/span>)<\/span>\n<\/code><\/pre>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/ba06323a3c6846631dec7f6b690e3f35.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p><strong>\u5411GPT4All\u63d0\u95ee\u5173\u4e8e\u60a8\u7684\u6587\u6863\u7684\u95ee\u9898<\/strong><\/p>\n<p>\u73b0\u5728\u6211\u4eec\u5230\u4e86\u8fd9\u4e00\u6b65\u3002\u6211\u4eec\u6709\u4e86\u6211\u4eec\u7684\u7d22\u5f15\uff0c\u6211\u4eec\u53ef\u4ee5\u52a0\u8f7d\u5b83\uff0c\u5e76\u4f7f\u7528\u4e00\u4e2a\u63d0\u793a\u6a21\u677f\u5411GPT4All\u63d0\u95ee\u3002\u6211\u4eec\u5148\u4ece\u4e00\u4e2a\u786c\u7f16\u7801\u7684\u95ee\u9898\u5f00\u59cb\uff0c\u7136\u540e\u5faa\u73af\u904d\u5386\u6211\u4eec\u7684\u8f93\u5165\u95ee\u9898\u3002<\/p>\n<p>\u5c06\u4ee5\u4e0b\u4ee3\u7801\u653e\u5165\u4e00\u4e2a\u540d\u4e3a db_loading.py \u7684 Python \u6587\u4ef6\u4e2d\uff0c\u5e76\u5728\u7ec8\u7aef\u4e2d\u4f7f\u7528 python3 db_loading.py \u547d\u4ee4\u8fd0\u884c\u5b83:<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-keyword\">from<\/span> langchain <span class=\"hljs-keyword\">import<\/span> PromptTemplate, LLMChain<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><span class=\"hljs-keyword\">from<\/span> langchain.llms <span class=\"hljs-keyword\">import<\/span> GPT4All<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.base <span class=\"hljs-keyword\">import<\/span> CallbackManager<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-keyword\">from<\/span> langchain.callbacks.streaming_stdout <span class=\"hljs-keyword\">import<\/span> StreamingStdOutCallbackHandler<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\"><span class=\"hljs-comment\"># function for loading only TXT files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> TextLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\"><span class=\"hljs-comment\"># text splitter for create chunks<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\"><span class=\"hljs-keyword\">from<\/span> langchain.text_splitter <span class=\"hljs-keyword\">import<\/span> RecursiveCharacterTextSplitter<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\"><span class=\"hljs-comment\"># to be able to load the pdf files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> UnstructuredPDFLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> PyPDFLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\"><span class=\"hljs-keyword\">from<\/span> langchain.document_loaders <span class=\"hljs-keyword\">import<\/span> DirectoryLoader<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\"><span class=\"hljs-comment\"># Vector Store Index to create our database about our knowledge<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\"><span class=\"hljs-keyword\">from<\/span> langchain.indexes <span class=\"hljs-keyword\">import<\/span> VectorstoreIndexCreator<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><span class=\"hljs-comment\"># LLamaCpp embeddings from the Alpaca model<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\"><span class=\"hljs-keyword\">from<\/span> langchain.embeddings <span class=\"hljs-keyword\">import<\/span> LlamaCppEmbeddings<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\"><span class=\"hljs-comment\"># FAISS  library for similaarity search<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\"><span class=\"hljs-keyword\">from<\/span> langchain.vectorstores.faiss <span class=\"hljs-keyword\">import<\/span> FAISS<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\"><span class=\"hljs-keyword\">import<\/span> os  <span class=\"hljs-comment\">#for interaaction with the files<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\"><span class=\"hljs-keyword\">import<\/span> datetime<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"22\"><span class=\"hljs-comment\"># TEST FOR SIMILARITY SEARCH<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"23\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"24\"><span class=\"hljs-comment\"># assign the path for the 2 models GPT4All and Alpaca for the embeddings <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"25\">gpt4all_path = <span class=\"hljs-string\">'.\/models\/gpt4all-converted.bin'<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"26\">llama_path = <span class=\"hljs-string\">'.\/models\/ggml-model-q4_0.bin'<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"27\"><span class=\"hljs-comment\"># Calback manager for handling the calls with  the model<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"28\">callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"29\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"30\"><span class=\"hljs-comment\"># create the embedding object<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"31\">embeddings = LlamaCppEmbeddings(model_path=llama_path)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"32\"><span class=\"hljs-comment\"># create the GPT4All llm object<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"33\">llm = GPT4All(model=gpt4all_path, callback_manager=callback_manager, verbose=<span class=\"hljs-literal\">True<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"34\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"35\"><span class=\"hljs-comment\"># Split text <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"36\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">split_chunks<\/span>(<span class=\"hljs-params\">sources<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"37\">    chunks = []<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"38\">    splitter = RecursiveCharacterTextSplitter(chunk_size=<span class=\"hljs-number\">256<\/span>, chunk_overlap=<span class=\"hljs-number\">32<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"39\">    <span class=\"hljs-keyword\">for<\/span> chunk <span class=\"hljs-keyword\">in<\/span> splitter.split_documents(sources):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"40\">        chunks.append(chunk)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"41\">    <span class=\"hljs-keyword\">return<\/span> chunks<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"42\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"43\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"44\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">create_index<\/span>(<span class=\"hljs-params\">chunks<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"45\">    texts = [doc.page_content <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> chunks]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"46\">    metadatas = [doc.metadata <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> chunks]<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"47\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"48\">    search_index = FAISS.from_texts(texts, embeddings, metadatas=metadatas)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"49\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"50\">    <span class=\"hljs-keyword\">return<\/span> search_index<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"51\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"52\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"53\"><span class=\"hljs-keyword\">def<\/span> <span class=\"hljs-title function_\">similarity_search<\/span>(<span class=\"hljs-params\">query, index<\/span>):<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"54\">    <span class=\"hljs-comment\"># k is the number of similarity searched that matches the query<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"55\">    <span class=\"hljs-comment\"># default is 4<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"56\">    matched_docs = index.similarity_search(query, k=<span class=\"hljs-number\">3<\/span>) <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"57\">    sources = []<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"58\">    <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> matched_docs:<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"59\">        sources.append(<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"60\">            {<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"61\">                <span class=\"hljs-string\">\"page_content\"<\/span>: doc.page_content,<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"62\">                <span class=\"hljs-string\">\"metadata\"<\/span>: doc.metadata,<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"63\">            }<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"64\">        )<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"65\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"66\">    <span class=\"hljs-keyword\">return<\/span> matched_docs, sources<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"67\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"68\"><span class=\"hljs-comment\"># Load our local index vector db<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"69\">index = FAISS.load_local(<span class=\"hljs-string\">\"my_faiss_index\"<\/span>, embeddings)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"70\"><span class=\"hljs-comment\"># Hardcoded question<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"71\">query = <span class=\"hljs-string\">\"What is a PLC and what is the difference with a PC\"<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"72\">docs = index.similarity_search(query)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"73\"><span class=\"hljs-comment\"># Get the matches best 3 results - defined in the function k=3<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"74\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"The question is: <span class=\"hljs-subst\">{query}<\/span>\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"75\"><span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">\"Here the result of the semantic search on the index, without GPT4All..\"<\/span>)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"76\"><span class=\"hljs-built_in\">print<\/span>(docs[<span class=\"hljs-number\">0<\/span>])<\/span>\n<\/code><\/pre>\n<p>\u6253\u5370\u7684\u6587\u672c\u662f\u4e0e\u67e5\u8be2\u6700\u5339\u914d\u7684\u524d3\u4e2a\u6765\u6e90\u7684\u5217\u8868\uff0c\u8fd8\u63d0\u4f9b\u4e86\u6587\u6863\u540d\u79f0\u548c\u9875\u7801\u3002<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/2fdeb939a1008ccdd851522e21dd0456.png\" alt=\"GPT4All\uff1aChatGPT\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72,\u7ec8\u751f\u514d\u8d39\" \/><\/figure>\n<\/p>\n<p>\u73b0\u5728\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u76f8\u4f3c\u6027\u641c\u7d22\u4f5c\u4e3a\u67e5\u8be2\u7684\u4e0a\u4e0b\u6587\uff0c\u4f7f\u7528\u63d0\u793a\u6a21\u677f\u3002\u5728\u8fd93\u4e2a\u51fd\u6570\u4e4b\u540e\uff0c\u53ea\u9700\u5c06\u6240\u6709\u4ee3\u7801\u66ff\u6362\u4e3a\u4ee5\u4e0b\u5185\u5bb9\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\"><span class=\"hljs-comment\"># Load our local index vector db<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\">index = FAISS.load_local(<span class=\"hljs-string\">\"my_faiss_index\"<\/span>, embeddings)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-comment\"># create the prompt template<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">template = <span class=\"hljs-string\">\"\"\"<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">Please use the following context to answer questions.<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\">Context: {context}<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">---<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\">Question: {question}<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\">Answer: Let's think step by step.\"\"\"<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\"><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\"><span class=\"hljs-comment\"># Hardcoded question<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\">question = <span class=\"hljs-string\">\"What is a PLC and what is the difference with a PC\"<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\">matched_docs, sources = similarity_search(question, index)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\"><span class=\"hljs-comment\"># Creating the context<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\">context = <span class=\"hljs-string\">\"n\"<\/span>.join([doc.page_content <span class=\"hljs-keyword\">for<\/span> doc <span class=\"hljs-keyword\">in<\/span> matched_docs])<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\"><span class=\"hljs-comment\"># instantiating the prompt template and the GPT4All chain<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\">prompt = PromptTemplate(template=template, input_variables=[<span class=\"hljs-string\">\"context\"<\/span>, <span class=\"hljs-string\">\"question\"<\/span>]).partial(context=context)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\">llm_chain = LLMChain(prompt=prompt, llm=llm)<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\"><span class=\"hljs-comment\"># Print the result<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\"><span class=\"hljs-built_in\">print<\/span>(llm_chain.run(question))<\/span>\n<\/code><\/pre>\n<p>\u8fd0\u884c\u4e4b\u540e\u4f60\u4f1a\u5f97\u5230\u4e00\u4e2a\u8fd9\u6837\u7684\u7ed3\u679c\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">vbnet<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-vbnet code-block-extension-codeShowNum\" lang=\"vbnet\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">Please use the following context <span class=\"hljs-keyword\">to<\/span> answer questions.<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"2\"><span class=\"hljs-symbol\">Context:<\/span> <span class=\"hljs-number\">1<\/span>.What <span class=\"hljs-built_in\">is<\/span> a PLC<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"3\"><span class=\"hljs-number\">2<\/span>.<span class=\"hljs-keyword\">Where<\/span> <span class=\"hljs-built_in\">and<\/span> Why it <span class=\"hljs-built_in\">is<\/span> used<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"4\"><span class=\"hljs-number\">3<\/span>.How a PLC <span class=\"hljs-built_in\">is<\/span> different <span class=\"hljs-keyword\">from<\/span> a PC<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"5\">PLC <span class=\"hljs-built_in\">is<\/span> especially important <span class=\"hljs-keyword\">in<\/span> industries <span class=\"hljs-keyword\">where<\/span> safety <span class=\"hljs-built_in\">and<\/span> reliability are<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"6\">critical, such <span class=\"hljs-keyword\">as<\/span> manufacturing plants, chemical plants, <span class=\"hljs-built_in\">and<\/span> power plants.<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"7\">How a PLC <span class=\"hljs-built_in\">is<\/span> different <span class=\"hljs-keyword\">from<\/span> a PC<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"8\">Because a PLC <span class=\"hljs-built_in\">is<\/span> a specialized computer used <span class=\"hljs-keyword\">in<\/span> industrial <span class=\"hljs-built_in\">and<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"9\">manufacturing applications <span class=\"hljs-keyword\">to<\/span> control machinery <span class=\"hljs-built_in\">and<\/span> processes.,the<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"10\">hardware components <span class=\"hljs-keyword\">of<\/span> a typical PLC must be able <span class=\"hljs-keyword\">to<\/span> interact <span class=\"hljs-keyword\">with<\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"11\">industrial device. So a typical PLC hardware include:<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"12\">---<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"13\"><span class=\"hljs-symbol\">Question:<\/span> What <span class=\"hljs-built_in\">is<\/span> a PLC <span class=\"hljs-built_in\">and<\/span> what <span class=\"hljs-built_in\">is<\/span> the difference <span class=\"hljs-keyword\">with<\/span> a PC<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"14\"><span class=\"hljs-symbol\">Answer:<\/span> <span class=\"hljs-keyword\">Let<\/span><span class=\"hljs-comment\">'s think step by step. 1) A Programmable Logic Controller (PLC), <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"15\">also called Industrial Control System <span class=\"hljs-built_in\">or<\/span> ICS, refers <span class=\"hljs-keyword\">to<\/span> an industrial computer <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"16\">that controls various automated processes such <span class=\"hljs-keyword\">as<\/span> manufacturing <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"17\">machines\/<span class=\"hljs-keyword\">assembly<\/span> lines etcetera through sensors <span class=\"hljs-built_in\">and<\/span> actuators connected <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"18\"><span class=\"hljs-keyword\">with<\/span> it via inputs &amp; outputs. It <span class=\"hljs-built_in\">is<\/span> a form <span class=\"hljs-keyword\">of<\/span> digital computers which has <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"19\">the ability <span class=\"hljs-keyword\">for<\/span> multiple instruction execution (MIE), built-<span class=\"hljs-keyword\">in<\/span> memory <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"20\">registers used <span class=\"hljs-keyword\">by<\/span> software routines, Input Output <span class=\"hljs-keyword\">interface<\/span> cards(IOC) <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"21\"><span class=\"hljs-keyword\">to<\/span> communicate <span class=\"hljs-keyword\">with<\/span> other devices electronically\/digitally over networks <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"22\"><span class=\"hljs-built_in\">or<\/span> buses etcetera<\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"23\"><span class=\"hljs-number\">2<\/span>). A Programmable Logic Controller <span class=\"hljs-built_in\">is<\/span> widely utilized <span class=\"hljs-keyword\">in<\/span> industrial <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"24\">automation <span class=\"hljs-keyword\">as<\/span> it has the ability <span class=\"hljs-keyword\">for<\/span> more than one instruction execution. <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"25\">It can perform tasks automatically <span class=\"hljs-built_in\">and<\/span> programmed instructions, which allows <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"26\">it <span class=\"hljs-keyword\">to<\/span> carry out complex operations that are beyond a <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"27\">Personal Computer (PC) capacity. So an ICS\/PLC contains built-<span class=\"hljs-keyword\">in<\/span> memory <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"28\">registers used <span class=\"hljs-keyword\">by<\/span> software routines <span class=\"hljs-built_in\">or<\/span> firmware codes etcetera but <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"29\">PC doesn<span class=\"hljs-comment\">'t contain them so they need external interfaces such as <\/span><\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"30\">hard disks drives(HDD), USB ports, serial <span class=\"hljs-built_in\">and<\/span> parallel <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"31\">communication protocols <span class=\"hljs-keyword\">to<\/span> store data <span class=\"hljs-keyword\">for<\/span> further analysis <span class=\"hljs-built_in\">or<\/span> <\/span>\n<span class=\"code-block-extension-codeLine\" data-line-num=\"32\">report generation.<\/span>\n<\/code><\/pre>\n<p>\u5982\u679c\u4f60\u5e0c\u671b\u4f7f\u7528\u7528\u6237\u8f93\u5165\u6765\u66ff\u6362\u95ee\u9898\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">question = <span class=\"hljs-string\">\"What is a PLC and what is the difference with a PC\"<\/span><\/span>\n<\/code><\/pre>\n<p>\u5c31\u50cf\u4e0b\u9762\u8fd9\u6837\uff1a<\/p>\n<pre><\/div><div class=\"code-block-extension-headerRight\"><span class=\"code-block-extension-lang\">python<\/span><div class=\"code-block-extension-copyCodeBtn\">\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div><code class=\"hljs language-python code-block-extension-codeShowNum\" lang=\"python\"><span class=\"code-block-extension-codeLine\" data-line-num=\"1\">question = <span class=\"hljs-built_in\">input<\/span>(<span class=\"hljs-string\">\"Your question: \"<\/span>)<\/span>\n<\/code><\/pre>\n<p>&nbsp;<\/p>\n<h1 data-id=\"heading-6\">\u7ed3\u8bba<\/h1>\n<p>\u73b0\u5728\u662f\u4f60\u8fdb\u884c\u5b9e\u9a8c\u7684\u65f6\u5019\u4e86\u3002\u5bf9\u4e0e\u4f60\u7684\u6587\u6863\u76f8\u5173\u7684\u6240\u6709\u4e3b\u9898\u63d0\u51fa\u4e0d\u540c\u7684\u95ee\u9898\uff0c\u5e76\u89c2\u5bdf\u7ed3\u679c\u3002\u5728 Prompt \u548c\u6a21\u677f\u65b9\u9762\uff0c\u80af\u5b9a\u8fd8\u6709\u5f88\u5927\u7684\u6539\u8fdb\u7a7a\u95f4\uff1a\u4f60\u53ef\u4ee5\u5728\u8fd9\u91cc\u627e\u5230\u4e00\u4e9b\u7075\u611f<a href=\"https:\/\/link.juejin.cn?target=https%3A%2F%2Fdocs.langchain.com%2Fdocs%2F%25E3%2580%2582\" target=\"_blank\" title=\"https:\/\/docs.langchain.com\/docs\/%E3%80%82\" ref=\"nofollow noopener noreferrer\" rel=\"noopener\">docs.langchain.com\/docs\/\u3002<\/a>\n\u800c\u4e14 Langchain \u7684\u6587\u6863\u771f\u7684\u5f88\u68d2\uff08\u6211\u90fd\u80fd\u591f\u7406\u89e3\uff01\uff09\u3002<\/p>\n<p>\u3010\u5fae\u4fe1\u516c\u4f17\u53f7\uff1aH5\u5f00\u8bb2\u5566\u3011<\/p>\n<p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/www.nicekj.com\/wp-content\/uploads\/replace\/9b8dbc8759d0a4ca8ad342b1a350e2a7.png\" alt=\"qrcode_for_gh_af6d0ab2a309_258.jpg\" \/><\/figure>\n<\/p>","protected":false},"excerpt":{"rendered":"<p>\u5728\u672c\u6587\u4e2d\uff0c\u6211\u4eec\u5c06\u5b66\u4e60\u5982\u4f55\u5728\u4ec5\u4f7f\u7528CPU\u7684\u8ba1\u7b97\u673a\u4e0a\u90e8\u7f72\u548c\u4f7f\u7528GPT4All\u6a21\u578b\u5e76\u5b66\u4e60\u5982\u4f55\u4f7f\u7528Python\u4e0e\u6211\u4eec\u7684\u6587\u6863\u8fdb\u884c\u4ea4\u4e92\u3002<\/p>\n","protected":false},"author":1,"featured_media":1292,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_title":"GPTAll | python | import - \u5948\u65af\u79d1\u6280\u793e\u533a","rank_math_description":"\u5948\u65af\u79d1\u6280\u793e\u533a\u4e13\u6ce8\u4e8eGPTAll\u3001python\u3001import\u3001question\u3001\u4ee3\u7801\uff0c\u63d0\u4f9b\u4e13\u4e1a\u7684\u5206\u4eab\u548c\u8ba8\u8bba\u5e73\u53f0\u3002","rank_math_focus_keyword":"GPTAll,python,import,question,\u4ee3\u7801","views":"41","footnotes":""},"categories":[3],"tags":[128,129,136,126,127],"collection":[],"class_list":["post-1157","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-fenlei2","tag-128","tag-129","tag-136","tag-gpt","tag-ai"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1157","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/comments?post=1157"}],"version-history":[{"count":0,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1157\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media\/1292"}],"wp:attachment":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media?parent=1157"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/categories?post=1157"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/tags?post=1157"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/collection?post=1157"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}