{"id":394,"date":"2026-03-23T05:49:55","date_gmt":"2026-03-23T09:49:55","guid":{"rendered":"https:\/\/mliaomcx.com\/?p=394"},"modified":"2026-03-29T09:49:47","modified_gmt":"2026-03-29T13:49:47","slug":"%e3%80%8a%e9%9b%b6%e6%88%90%e6%9c%ac%e5%ae%9e%e7%8e%b0%e5%a4%a7%e6%a8%a1%e5%9e%8b-api-%e8%87%aa%e7%94%b1%ef%bc%9agoogle-colab-%e9%83%a8%e7%bd%b2-ollama-vllm-%e7%bb%88%e6%9e%81%e6%8c%87%e5%8d%97","status":"publish","type":"post","link":"https:\/\/mliaomcx.com\/en\/%e3%80%8a%e9%9b%b6%e6%88%90%e6%9c%ac%e5%ae%9e%e7%8e%b0%e5%a4%a7%e6%a8%a1%e5%9e%8b-api-%e8%87%aa%e7%94%b1%ef%bc%9agoogle-colab-%e9%83%a8%e7%bd%b2-ollama-vllm-%e7%bb%88%e6%9e%81%e6%8c%87%e5%8d%97\/","title":{"rendered":"\u300a\u96f6\u6210\u672c\u5b9e\u73b0\u5927\u6a21\u578b API \u81ea\u7531\uff1aGoogle Colab \u90e8\u7f72 Ollama + vLLM \u7ec8\u6781\u6307\u5357\u300b"},"content":{"rendered":"\n<p><strong>\u6458\u8981 (Meta Description):<\/strong> \u538c\u5026\u4e86\u9ad8\u6602\u7684 API \u8d26\u5355\uff1f\u672c\u6587\u624b\u628a\u624b\u6559\u4f60\u5229\u7528 Google Colab \u514d\u8d39 T4 GPU\uff0c\u5728\u4e91\u7aef\u96f6\u6210\u672c\u90e8\u7f72 Ollama \u4e0e vLLM \u9ad8\u6027\u80fd\u5927\u6a21\u578b\u3002\u914d\u5408 Cloudflare Tunnel \u5185\u7f51\u7a7f\u900f\u3001Cherry Studio \u5ba2\u6237\u7aef\uff0c\u5b8c\u7f8e\u652f\u6491 OpenClaw \u81ea\u52a8\u5316\u5168\u80fd Agent\uff0c\u5f7b\u5e95\u5b9e\u73b0 OpenAI \u517c\u5bb9 API \u81ea\u7531\u3002<\/p>\n\n\n\n<p>\u5e73\u65f6\u5f00\u53d1\u5404\u7c7b AI \u5e94\u7528\u3001\u8dd1\u672c\u5730 Agent \u6d41\u7a0b\uff0c\u6216\u8005\u91cd\u5ea6\u4f7f\u7528\u6c89\u6d78\u5f0f\u7ffb\u8bd1\uff0c\u6700\u75db\u7684\u5f80\u5f80\u4e0d\u662f\u4ee3\u7801 bug\uff0c\u800c\u662f\u773c\u770b\u7740\u5404\u5927\u5382 API \u989d\u5ea6\u75af\u72c2\u71c3\u70e7\u7684\u8d26\u5355\u3002<\/p>\n\n\n\n<p>\u5f88\u591a\u670b\u53cb\u4f1a\u5c1d\u8bd5\u5728\u672c\u5730\u8dd1\u6a21\u578b\uff0c\u4f46\u5982\u679c\u4f60\u624b\u91cc\u53ea\u6709\u4e00\u53f0\u642d\u8f7d RTX 4060 (8GB \u663e\u5b58) \u8fd9\u6837\u7684\u4e3b\u6d41\u7b14\u8bb0\u672c\uff0c\u60f3\u8981\u517c\u987e\u8d85\u957f\u4e0a\u4e0b\u6587\u548c\u9ad8\u901f\u5e76\u53d1\uff0c\u663e\u5b58\u6839\u672c\u6349\u895f\u89c1\u8098\u3002\u4eca\u5929\uff0c\u6211\u4eec\u76f4\u63a5\u8f6c\u6362\u601d\u8def\uff1a<strong>\u767d\u5ad6 Google Colab \u514d\u8d39\u63d0\u4f9b\u7684 16GB T4 GPU\uff0c\u5728\u4e91\u7aef\u90e8\u7f72\u5f00\u6e90\u5927\u6a21\u578b\uff0c\u5f7b\u5e95\u5b9e\u73b0 API \u81ea\u7531\u3002<\/strong><\/p>\n\n\n\n<p>\u672c\u6559\u7a0b\u5c06\u5e26\u4f60\u4ece\u96f6\u5f00\u59cb\uff0c\u5b9e\u64cd\u4e1a\u5185\u6700\u706b\u7684\u4e24\u5927\u63a8\u7406\u6846\u67b6\u2014\u2014Ollama \u4e0e vLLM\uff0c\u5e76\u6700\u7ec8\u6253\u901a\u786c\u6838\u81ea\u52a8\u5316\u5de5\u5177 OpenClaw\u3002<\/p>\n\n\n\n<p><strong>\ud83d\udc47 \u6559\u7a0b\u914d\u5957\u7684\u4e00\u952e\u8fd0\u884c\u811a\u672c\u5730\u5740\uff1a<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\ud83d\udc49 <strong>[ <a href=\"https:\/\/colab.research.google.com\/drive\/1JQvL11x3Kaue94tBsy9arAGXA1wWVsXi?usp=sharing\">Ollama \u90e8\u7f72 Colab \u5730\u5740<\/a>]<\/strong><\/li>\n\n\n\n<li>\ud83d\udc49 <strong>[ <a href=\"https:\/\/colab.research.google.com\/drive\/1aIcNMieaoacndHQ-kMpwu2KnfoyBq65h?usp=sharing\">vLLM + AWQ \u90e8\u7f72 Colab \u5730\u5740<\/a>]<\/strong><\/li>\n<\/ul>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e3a\u4ec0\u4e48\u662f Ollama \u548c vLLM\uff1f(\u6846\u67b6\u5bf9\u6bd4)<\/h2>\n\n\n\n<p>\u5728\u52a8\u624b\u524d\uff0c\u6211\u4eec\u9700\u8981\u660e\u786e\u4e24\u79cd\u4e3b\u6d41\u6846\u67b6\u7684\u5b9a\u4f4d\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Ollama (\u7075\u6d3b\u5168\u80fd\u738b)\uff1a<\/strong> \u90e8\u7f72\u6781\u5176\u7b80\u5355\uff0c\u652f\u6301 CPU\/GPU\uff0c\u4e0d\u6311\u786c\u4ef6\u3002\u975e\u5e38\u9002\u5408\u65e5\u5e38\u591a\u6a21\u6001\u6a21\u578b\u5207\u6362\u3001\u8f7b\u91cf\u7ea7\u5bf9\u8bdd\u3002\u4f46\u5728\u9ad8\u5e76\u53d1\u6781\u9650\u541e\u5410\u573a\u666f\u4e0b\uff0c\u6027\u80fd\u7565\u6709\u74f6\u9888\u3002<\/li>\n\n\n\n<li><strong>vLLM (\u6027\u80fd\u602a\u517d)\uff1a<\/strong> \u4e13\u4e3a\u52a0\u901f\u5927\u8bed\u8a00\u6a21\u578b\u63a8\u7406\u800c\u751f\u7684 GPU \u63a8\u7406\u5f15\u64ce\u3002\u6781\u901f\u5e76\u53d1\uff0c\u662f\u5de5\u4e1a\u7ea7\u3001\u56e2\u961f\u7ea7 API \u5206\u53d1\u7684\u9996\u9009\u3002\u7f3a\u70b9\u662f\u5bf9\u786c\u4ef6\u548c\u73af\u5883\u8981\u6c42\u7a0d\u9ad8\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u6211\u4eec\u5c06\u5229\u7528 Colab \u7684 T4 \u663e\u5361\uff0c\u628a\u8fd9\u4e24\u5957\u65b9\u6848\u90fd\u73a9\u5230\u6781\u81f4\u3002\u6253\u5f00\u4e0a\u65b9\u63d0\u4f9b\u7684 Colab \u94fe\u63a5\u540e\uff0c\u70b9\u51fb\u9876\u90e8\u83dc\u5355\u680f\u7684 <strong>\u4ee3\u7801\u6267\u884c\u7a0b\u5e8f -&gt; \u66f4\u6539\u4ee3\u7801\u6267\u884c\u7a0b\u5e8f\u7c7b\u578b<\/strong>\uff0c\u786e\u4fdd\u201c\u786c\u4ef6\u52a0\u901f\u5668\u201d\u9009\u62e9\u7684\u662f <strong>T4 GPU<\/strong>\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h2 class=\"wp-block-heading\">\u65b9\u6848 A\uff1aOllama \u591a\u6a21\u6001\u5168\u80fd\u90e8\u7f72<\/h2>\n\n\n\n<p>\ud83d\udc49 <strong>[ <a href=\"https:\/\/colab.research.google.com\/drive\/1JQvL11x3Kaue94tBsy9arAGXA1wWVsXi?usp=sharing\">Ollama \u90e8\u7f72 Colab \u5730\u5740<\/a>]<\/strong><\/p>\n\n\n\n<p><strong>\u9002\u7528\u573a\u666f\uff1a<\/strong> \u65e5\u5e38\u804a\u5929\u3001\u56fe\u7247\u5206\u6790\uff0c\u652f\u6301\u7eaf\u6587\u672c\u4e0e\u89c6\u89c9\u6a21\u578b\u65e0\u7f1d\u5207\u6362\u3002<\/p>\n\n\n\n<p>\u5728\u7b2c\u4e00\u5957\u65b9\u6848\u4e2d\uff0c\u6211\u4eec\u540c\u65f6\u62c9\u53d6\u4e24\u4e2a\u6a21\u578b\uff1a\u8d1f\u8d23\u5f3a\u608d\u7eaf\u6587\u672c\u903b\u8f91\u63a8\u7406\u7684 <code>Qwen3.5:9b<\/code>\uff0c\u4ee5\u53ca\u8d1f\u8d23\u5f3a\u5927\u7684\u591a\u6a21\u6001\u89c6\u89c9\u5206\u6790\u7684 <code>Qwen3-vl:8b<\/code>\u3002<\/p>\n\n\n\n<p><strong>\u26a0\ufe0f \u6572\u9ed1\u677f\uff0c\u786c\u6838\u907f\u5751\u6307\u5357\uff1a<\/strong> Qwen 3.5 \u662f 2026 \u5e74\u521a\u51fa\u7684\u5168\u65b0\u67b6\u6784\u3002\u5f88\u591a\u4eba\u7528\u7b2c\u4e09\u65b9\u8001\u5f15\u64ce\uff08\u5982\u65e7\u7248 llama-cpp-python\uff09\u90e8\u7f72\u65f6\uff0c\u7531\u4e8e\u5e95\u5c42\u4ee3\u7801\u8fd8\u6ca1\u5199\u8fdb qwen35 \u7684\u5b9a\u4e49\uff0c\u5f15\u64ce\u201c\u4e0d\u8ba4\u8bc6\u201d\u5b83\uff0c\u65e5\u5fd7\u91cc\u4f1a\u76f4\u63a5\u62a5\u51fa <code>unknown model architecture: 'qwen35'<\/code> \u7684\u81f4\u547d\u9519\u8bef\u5e76\u5f3a\u884c\u9000\u51fa\u3002 \u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\uff0c\u5728\u6211\u7684 Colab \u811a\u672c\u4e2d\uff0c\u76f4\u63a5\u4f7f\u7528\u4e86\u5b98\u65b9\u811a\u672c\u5b89\u88c5\u6700\u65b0\u7248 Ollama\uff0c\u5e95\u5c42\u517c\u5bb9\u6027\u62c9\u6ee1\uff0c\u5b8c\u7f8e\u907f\u5f00\u62a5\u9519\uff0c\u4e00\u628a\u8fc7\uff01<\/p>\n\n\n\n<p><strong>\u83b7\u53d6 API \u94fe\u63a5\uff1a<\/strong> Colab \u5728\u4e91\u7aef\u5185\u7f51\uff0c\u5343\u4e07\u522b\u7528\u4f1a\u5f39\u7a97\u62e6\u622a\u7684 localtunnel\uff01\u811a\u672c\u4e2d\u5185\u7f6e\u4e86\u4f01\u4e1a\u7ea7\u7684 <strong>Cloudflare Tunnel<\/strong> \u8fdb\u884c\u5185\u7f51\u7a7f\u900f\u3002\u8fd0\u884c\u540e\uff0c\u5728\u65e5\u5fd7\u91cc\u627e\u5230\u4ee5 <code>trycloudflare.com<\/code> \u7ed3\u5c3e\u7684\u516c\u7f51 URL\uff0c\u8fd9\u5c31\u662f\u4f60\u6781\u5176\u7a33\u5b9a\u4e14\u6ca1\u6709\u4efb\u4f55\u62e6\u622a\u7684\u4e13\u5c5e API \u5730\u5740\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h2 class=\"wp-block-heading\">\u65b9\u6848 B\uff1avLLM + AWQ \u6781\u9650\u6027\u80fd\u538b\u69a8<\/h2>\n\n\n\n<p>\ud83d\udc49 <strong>[<a href=\"https:\/\/colab.research.google.com\/drive\/1aIcNMieaoacndHQ-kMpwu2KnfoyBq65h?usp=sharing\">vLLM + AWQ \u90e8\u7f72 Colab \u5730\u5740<\/a>]<\/strong><\/p>\n\n\n\n<p><strong>\u9002\u7528\u573a\u666f\uff1a<\/strong> \u9700\u8981\u6781\u901f\u54cd\u5e94\u548c\u8d85\u5927\u4e0a\u4e0b\u6587\u7684\u81ea\u52a8\u5316 Agent \u4efb\u52a1\u3002<\/p>\n\n\n\n<p>\u5982\u679c\u4f60\u8ffd\u6c42\u6781\u81f4\u6253\u5b57\u901f\u5ea6\uff0c\u770b\u7b2c\u4e8c\u79cd\u65b9\u6848\uff1avLLM\u3002Colab \u7684 T4 \u663e\u5361\u53ea\u6709 16GB \u663e\u5b58\uff0c\u76f4\u63a5\u8dd1\u539f\u7248 7B \u6a21\u578b\uff0c\u663e\u5b58\u5403\u7d27\uff0cvLLM \u6839\u672c\u65e0\u6cd5\u53d1\u6325\u5e76\u53d1\u4f18\u52bf\u3002<\/p>\n\n\n\n<p><strong>\u6280\u672f\u6838\u5fc3\uff1a<\/strong> \u8fd9\u91cc\u9009\u7528 <code>Qwen2.5-7B-Instruct-AWQ<\/code> \u91cf\u5316\u6a21\u578b\u3002AWQ \u662f\u4e00\u79cd\u5148\u8fdb\u7684\u6a21\u578b\u538b\u7f29\u6280\u672f\uff0c\u80fd\u5728\u51e0\u4e4e\u4e0d\u635f\u5931\u667a\u5546\u7684\u524d\u63d0\u4e0b\uff0c\u628a\u9700\u8981 14GB \u663e\u5b58\u7684\u6a21\u578b\u538b\u7f29\u5230 5GB \u5c31\u80fd\u8dd1\u3002\u5269\u4e0b\u7684 11GB \u663e\u5b58\u5168\u90e8\u7559\u7ed9 vLLM \u53bb\u505a\u6781\u901f\u5e76\u53d1\uff08KV Cache\uff09\u3002\u8fd9\u5c31\u53eb\u5b8c\u7f8e\u538b\u69a8\u4e91\u7aef\u6027\u80fd\uff01<\/p>\n\n\n\n<p>\u540c\u6837\uff0c\u811a\u672c\u4f1a\u81ea\u52a8\u4f7f\u7528 Cloudflare Tunnel \u7a7f\u900f\u9ed8\u8ba4\u7684 8000 \u7aef\u53e3\uff0c\u8bf7\u63d0\u53d6\u65e5\u5fd7\u4e2d\u7684\u516c\u7f51 URL \u5907\u7528\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h2 class=\"wp-block-heading\">\u684c\u9762\u7aef\u5b9e\u6218\u4e00\uff1aCherry Studio \u8be6\u7ec6\u914d\u7f6e\u6307\u5357<\/h2>\n\n\n\n<p>\u62ff\u5230\u6781\u5176\u7a33\u5b9a\u7684 API \u63a5\u53e3\u540e\uff0c\u65e5\u5e38\u4f7f\u7528\u5f3a\u70c8\u63a8\u8350\u5f00\u6e90\u795e\u5668 <strong>Cherry Studio<\/strong>\u3002<\/p>\n\n\n\n<p>cherr studio\u7684\u4e0b\u8f7d\u5730\u5740\uff1a<a href=\"https:\/\/www.cherry-ai.com\/download\">https:\/\/www.cherry-ai.com\/download<\/a><\/p>\n\n\n\n<p><strong>\u914d\u7f6e\u6b65\u9aa4\uff1a<\/strong><\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li>\u542f\u52a8 Cherry Studio\uff0c\u8fdb\u5165\u201c\u8bbe\u7f6e\u201d\uff0c\u5728\u6a21\u578b\u670d\u52a1\u5546\u4e2d\u9009\u62e9 <strong>OpenAI<\/strong>\uff08\u6211\u4eec\u7684\u670d\u52a1\u5b8c\u7f8e\u517c\u5bb9 OpenAI \u63a5\u53e3\uff09\u3002<\/li>\n\n\n\n<li><strong>API \u5bc6\u94a5\uff1a<\/strong>\u7559\u7a7a<\/li>\n\n\n\n<li><strong>API \u5730\u5740\uff1a<\/strong> \u7c98\u8d34\u4f60\u83b7\u53d6\u7684 Cloudflare URL\uff0c<strong>\u6ce8\u610f\u5fc5\u987b\u5728\u7f51\u5740\u672b\u5c3e\u52a0\u4e0a <code>\/v1<\/code><\/strong>\u3002<\/li>\n\n\n\n<li>\u624b\u52a8\u6dfb\u52a0\u6a21\u578b\u540d\u79f0\uff08\u5982 <code>qwen3.5:9b<\/code>\u3001qwen3.5:9b\u6216 <code>Qwen\/Qwen2.5-7B-Instruct-AWQ<\/code>\uff09\u3002<\/li>\n<\/ol>\n\n\n\n<p>Cherry Studio \u4e0d\u4ec5\u652f\u6301\u591a\u6a21\u578b\u65e0\u7f1d\u5207\u6362\uff0c\u8fd8\u53ef\u4ee5\u8bbe\u5b9a\u201c\u8d44\u6df1\u67b6\u6784\u5e08\u201d\u7b49 System Prompt\u3002\u6700\u738b\u70b8\u7684\u662f\u5b83<strong>\u81ea\u5e26\u514d\u8d39\u8054\u7f51\u641c\u7d22<\/strong>\uff0c\u914d\u5408\u6211\u4eec\u7684\u4e91\u7aef\u5927\u6a21\u578b\uff0c\u81ea\u52a8\u83b7\u53d6\u6700\u65b0\u8d44\u8baf\u5e76\u603b\u7ed3\uff0c\u4f53\u9a8c\u5b8c\u5168\u4e0d\u8f93\u5927\u5382\u7684\u4ed8\u8d39 Plus \u4f1a\u5458\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h2 class=\"wp-block-heading\">\u7ed3\u8bed<\/h2>\n\n\n\n<p>\u4ece Ollama \u7684\u591a\u6a21\u6001\u90e8\u7f72\uff0c\u5230\u89c4\u907f Qwen 3.5 \u7684\u67b6\u6784\u517c\u5bb9\u5751\uff1b\u4ece vLLM + AWQ \u6781\u9650\u538b\u69a8\u663e\u5b58\uff0c\u5230 Cloudflare \u89e3\u51b3\u4f01\u4e1a\u7ea7\u5185\u7f51\u7a7f\u900f\uff0c\u8fd9\u5957\u7ec4\u5408\u62f3\u5c06 Google Colab \u7684\u767d\u5ad6\u4ef7\u503c\u53d1\u6325\u5230\u4e86\u6781\u81f4\u3002<\/p>\n\n\n\n<p>\u5982\u679c\u5728\u90e8\u7f72\u6216\u914d\u7f6e\u8fc7\u7a0b\u4e2d\u9047\u5230\u4efb\u4f55\u62a5\u9519\uff0c\u6b22\u8fce\u5728\u8bc4\u8bba\u533a\u7559\u8a00\u8ba8\u8bba\u3002\u60f3\u4e86\u89e3\u66f4\u591a\u5173\u4e8e\u5927\u6a21\u578b\u90e8\u7f72\u3001\u7f51\u7edc\u7a7f\u900f\u7684\u786c\u6838\u6298\u817e\u73a9\u6cd5\uff0c\u8bb0\u5f97\u5728 YouTube \u8ba2\u9605\u6211\u7684\u9891\u9053 <strong>@mcxmliao<\/strong> (Mliaomcx)\uff0c\u6211\u4eec\u4e0b\u671f\u518d\u89c1\uff01<\/p>\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6458\u8981 (Meta Descri&#8230;<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-394","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/posts\/394","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/comments?post=394"}],"version-history":[{"count":3,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/posts\/394\/revisions"}],"predecessor-version":[{"id":399,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/posts\/394\/revisions\/399"}],"wp:attachment":[{"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/media?parent=394"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/categories?post=394"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mliaomcx.com\/en\/wp-json\/wp\/v2\/tags?post=394"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}