{"id":61,"date":"2026-04-06T16:52:46","date_gmt":"2026-04-06T08:52:46","guid":{"rendered":"https:\/\/blog.dinonode.org\/?p=61"},"modified":"2026-04-06T16:54:41","modified_gmt":"2026-04-06T08:54:41","slug":"ollama-%e6%9c%ac%e5%9c%b0%e6%a8%a1%e5%9e%8b%e9%83%a8%e7%bd%b2%e5%85%a8%e6%8c%87%e5%8d%97%ef%bc%9a%e5%ae%89%e8%a3%85%e3%80%81%e4%bd%bf%e7%94%a8%e4%b8%8e%e6%b3%a8%e6%84%8f%e4%ba%8b%e9%a1%b9","status":"publish","type":"post","link":"https:\/\/blog.dinonode.org\/index.php\/2026\/04\/06\/ollama-%e6%9c%ac%e5%9c%b0%e6%a8%a1%e5%9e%8b%e9%83%a8%e7%bd%b2%e5%85%a8%e6%8c%87%e5%8d%97%ef%bc%9a%e5%ae%89%e8%a3%85%e3%80%81%e4%bd%bf%e7%94%a8%e4%b8%8e%e6%b3%a8%e6%84%8f%e4%ba%8b%e9%a1%b9\/","title":{"rendered":"Ollama \u672c\u5730\u6a21\u578b\u90e8\u7f72\u5168\u6307\u5357\uff1a\u5b89\u88c5\u3001\u4f7f\u7528\u4e0e\u6ce8\u610f\u4e8b\u9879"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">\u5728\u672c\u5730\u79c1\u6709\u5316\u90e8\u7f72\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u65e2\u80fd\u4fdd\u969c\u6570\u636e\u9690\u79c1\uff0c\u4e5f\u80fd\u8131\u79bb\u7f51\u7edc\u9650\u5236\u7a33\u5b9a\u4f7f\u7528\u3002Ollama \u662f\u4e00\u6b3e\u5f00\u6e90\u3001\u8de8\u5e73\u53f0\u7684\u5927\u6a21\u578b\u8fd0\u884c\u5de5\u5177\uff0c\u652f\u6301\u4e00\u952e\u90e8\u7f72 Llama 3\u3001Qwen\u3001DeepSeek \u7b49\u4e3b\u6d41\u6a21\u578b\uff0c\u547d\u4ee4\u884c\u64cd\u4f5c\u7b80\u6d01\uff0c\u65e0\u9700\u590d\u6742\u73af\u5883\u914d\u7f6e\uff0c\u662f\u4e2a\u4eba\u4e0e\u5c0f\u578b\u56e2\u961f\u672c\u5730\u90e8\u7f72\u7684\u4f18\u9009\u65b9\u6848\u3002\u672c\u6587\u5c06\u5b8c\u6574\u8bb2\u89e3 Ollama \u7684\u90e8\u7f72\u6d41\u7a0b\u3001\u57fa\u7840\u4f7f\u7528\uff0c\u4ee5\u53ca\u90e8\u7f72\u4e2d\u5fc5\u987b\u5173\u6ce8\u7684\u6838\u5fc3\u6ce8\u610f\u4e8b\u9879\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e00\u3001Ollama \u6838\u5fc3\u4f18\u52bf<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8de8\u5e73\u53f0\u517c\u5bb9\uff1a\u652f\u6301 Windows\u3001macOS\u3001Linux\uff0c\u4e5f\u53ef\u901a\u8fc7 Docker \u5bb9\u5668\u8fd0\u884c<\/li>\n\n\n\n<li>\u6781\u7b80\u90e8\u7f72\uff1a\u4e00\u884c\u547d\u4ee4\u5b8c\u6210\u5b89\u88c5\u4e0e\u6a21\u578b\u62c9\u53d6\uff0c\u65e0\u9700\u624b\u52a8\u914d\u7f6e\u4f9d\u8d56<\/li>\n\n\n\n<li>\u6a21\u578b\u751f\u6001\u5b8c\u5584\uff1a\u5185\u7f6e\u5b98\u65b9\u6a21\u578b\u5e93\uff0c\u652f\u6301 GGUF \u683c\u5f0f\u6a21\u578b\u5bfc\u5165<\/li>\n\n\n\n<li>\u672c\u5730\u9690\u79c1\u5b89\u5168\uff1a\u6240\u6709\u63a8\u7406\u4e0e\u6570\u636e\u5b58\u50a8\u5728\u672c\u5730\uff0c\u4e0d\u4e0a\u4f20\u4e91\u7aef<\/li>\n\n\n\n<li>\u5f00\u653e API\uff1a\u517c\u5bb9 OpenAI \u683c\u5f0f\u63a5\u53e3\uff0c\u53ef\u5feb\u901f\u5bf9\u63a5\u524d\u7aef\u3001\u5e94\u7528\u7a0b\u5e8f<\/li>\n\n\n\n<li>\u79c1\u6709\u5316\u5b9a\u5236\uff1a\u652f\u6301\u79c1\u6709 API \u90e8\u7f72\u3001\u9274\u6743\u914d\u7f6e\u3001\u5c40\u57df\u7f51 \/ \u5185\u7f51\u8bbf\u95ee<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e8c\u3001\u90e8\u7f72\u524d\u51c6\u5907<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">1. \u786c\u4ef6\u914d\u7f6e\u8981\u6c42<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Ollama \u7684\u786c\u4ef6\u9700\u6c42\u4e3b\u8981\u53d6\u51b3\u4e8e\u8fd0\u884c\u6a21\u578b\u7684\u53c2\u6570\u91cf\uff0c\u6838\u5fc3\u914d\u7f6e\u5efa\u8bae\u5982\u4e0b\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8f7b\u91cf\u6a21\u578b\uff081.5B-7B\uff09\uff1a\u6700\u4f4e 8GB \u5185\u5b58\uff0c\u63a8\u8350 16GB \u5185\u5b58<\/li>\n\n\n\n<li>\u4e2d\u7b49\u6a21\u578b\uff0813B\uff09\uff1a16GB \u5185\u5b58\u8d77\u6b65\uff0c\u63a8\u8350 32GB \u5185\u5b58\uff0c\u5e94\u4f7f\u7528 GPU \u52a0\u901f<\/li>\n\n\n\n<li>\u5927\u6a21\u578b\uff0833B \u53ca\u4ee5\u4e0a\uff09\uff1a32GB + \u5185\u5b58\uff0c\u663e\u5b58>=24G\uff0c\u4f18\u5148 NVIDIA\/AMD \u663e\u5361<\/li>\n\n\n\n<li>\u5b58\u50a8\uff1a\u9884\u7559 20GB \u4ee5\u4e0a\u53ef\u7528\u7a7a\u95f4\uff0c\u6a21\u578b\u6587\u4ef6\u5927\u5c0f\u7ea6 2GB-40GB \u4e0d\u7b49<\/li>\n\n\n\n<li>GPU\uff1aNVIDIA \u663e\u5361\u9700\u652f\u6301 CUDA\uff0cAMD \u652f\u6301 ROCm\uff0cApple Silicon \u539f\u751f\u52a0\u901f<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">2. GPU \u663e\u5b58\u8be6\u7ec6\u8981\u6c42<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">\u663e\u5b58\u5360\u7528\u4e0e<strong>\u6a21\u578b\u53c2\u6570\u91cf + \u91cf\u5316\u7cbe\u5ea6<\/strong>\u76f4\u63a5\u76f8\u5173\uff0c\u4e3a\u90e8\u7f72\u6838\u5fc3\u53c2\u8003\u6807\u51c6\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>7B \u6a21\u578b\uff08\u4e3b\u6d41\u63a8\u8350\uff09\n<ul class=\"wp-block-list\">\n<li>Q4_K_M\uff08\u91cf\u5316\uff09\uff1a<strong>\u663e\u5b58\u22654GB<\/strong>\uff08\u6700\u4f4e\u53ef\u7528\uff09\uff0c\u63a8\u8350<strong>6GB+<\/strong><\/li>\n\n\n\n<li>Q5_K_M\uff08\u91cf\u5316\uff09\uff1a<strong>\u663e\u5b58\u22656GB<\/strong><\/li>\n\n\n\n<li>FP16\uff08\u5168\u7cbe\u5ea6\uff09\uff1a<strong>\u663e\u5b58\u226513GB<\/strong><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>13B \u6a21\u578b\n<ul class=\"wp-block-list\">\n<li>Q4_K_M\uff08\u91cf\u5316\uff09\uff1a<strong>\u663e\u5b58\u22658GB<\/strong>\uff0c\u63a8\u8350<strong>10GB+<\/strong><\/li>\n\n\n\n<li>Q5_K_M\uff08\u91cf\u5316\uff09\uff1a<strong>\u663e\u5b58\u226510GB<\/strong><\/li>\n\n\n\n<li>FP16\uff08\u5168\u7cbe\u5ea6\uff09\uff1a<strong>\u663e\u5b58\u226526GB<\/strong><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>33B \u6a21\u578b\n<ul class=\"wp-block-list\">\n<li>Q4_K_M\uff08\u91cf\u5316\uff09\uff1a<strong>\u663e\u5b58\u226516GB<\/strong>\uff0c\u63a8\u8350<strong>20GB+<\/strong><\/li>\n\n\n\n<li>FP16\uff08\u5168\u7cbe\u5ea6\uff09\uff1a<strong>\u663e\u5b58\u226565GB<\/strong><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u901a\u7528\u89c4\u5219\uff1a\u65e0\u72ec\u7acb\u663e\u5361\u4ec5\u53ef\u8fd0\u884c CPU \u63a8\u7406\uff0c\u663e\u5b58\u4e0d\u8db3\u4f1a\u76f4\u63a5\u5bfc\u81f4\u6a21\u578b\u52a0\u8f7d\u5931\u8d25\u3001\u95ea\u9000\u3002<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">3. \u7cfb\u7edf\u8981\u6c42<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Windows\uff1aWindows 10 2004 \u7248\u672c\u53ca\u4ee5\u4e0a\u3001Windows 11<\/li>\n\n\n\n<li>macOS\uff1amacOS 11.0 \u53ca\u4ee5\u4e0a\uff0c\u652f\u6301 Apple Silicon \u4e0e Intel \u82af\u7247<\/li>\n\n\n\n<li>Linux\uff1aUbuntu 20.04+\u3001Debian 11+\u3001CentOS 8 + \u7b49\u4e3b\u6d41\u53d1\u884c\u7248<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">4. \u5176\u4ed6\u51c6\u5907<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u7f51\u7edc\uff1a\u9996\u6b21\u90e8\u7f72\u9700\u8054\u7f51\u4e0b\u8f7d\u5b89\u88c5\u5305\u4e0e\u6a21\u578b\uff0c\u79bb\u7ebf\u540e\u53ef\u6b63\u5e38\u4f7f\u7528<\/li>\n\n\n\n<li>\u6743\u9650\uff1aWindows \u9700\u7ba1\u7406\u5458\u6743\u9650\u5b89\u88c5\uff0cLinux\/macOS \u9700 sudo \u6743\u9650<\/li>\n\n\n\n<li>\u5b58\u50a8\u89c4\u5212\uff1a\u907f\u514d\u5c06\u6a21\u578b\u5b58\u50a8\u5728\u7cfb\u7edf\u76d8\uff0c\u9632\u6b62\u7a7a\u95f4\u4e0d\u8db3<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e09\u3001\u5168\u5e73\u53f0\u5b89\u88c5\u6b65\u9aa4<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">1. Windows \u5b89\u88c5<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u8bbf\u95ee Ollama \u5b98\u7f51\u4e0b\u8f7d Windows \u5b89\u88c5\u5305<\/li>\n\n\n\n<li>\u53cc\u51fb\u8fd0\u884c\u5b89\u88c5\u7a0b\u5e8f\uff0c\u81ea\u52a8\u914d\u7f6e\u73af\u5883\u53d8\u91cf\u4e0e\u540e\u53f0\u670d\u52a1<\/li>\n\n\n\n<li>\u5b89\u88c5\u5b8c\u6210\u540e\u91cd\u542f\u7ec8\u7aef\uff0c\u8f93\u5165<code>ollama --version<\/code>\u9a8c\u8bc1\u5b89\u88c5<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">2. macOS \u5b89\u88c5<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">\u6253\u5f00\u7ec8\u7aef\uff0c\u6267\u884c\u4e00\u952e\u5b89\u88c5\u811a\u672c\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>curl -fsSL https:\/\/ollama.com\/install.sh | sh\n<\/code><\/pre>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b89\u88c5\u5b8c\u6210\u540e\u8f93\u5165<code>ollama --version<\/code>\u9a8c\u8bc1\uff0cApple Silicon \u82af\u7247\u53ef\u539f\u751f\u542f\u7528 GPU \u52a0\u901f\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">3. Linux \u5b89\u88c5<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">\u7ec8\u7aef\u6267\u884c\u5b98\u65b9\u4e00\u952e\u811a\u672c\uff0c\u81ea\u52a8\u5b8c\u6210\u4e0b\u8f7d\u3001\u670d\u52a1\u914d\u7f6e\u4e0e\u542f\u52a8\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>curl -fsSL https:\/\/ollama.com\/install.sh | sh\n<\/code><\/pre>\n\n\n\n<p class=\"wp-block-paragraph\">\u9a8c\u8bc1\u547d\u4ee4\uff1a<code>ollama --version<\/code>\uff0c\u67e5\u770b\u7248\u672c\u53f7\u5373\u5b89\u88c5\u6210\u529f\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">4. Docker \u90e8\u7f72\uff08\u53ef\u9009\uff09<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">\u9002\u5408\u670d\u52a1\u5668\u4e0e\u5bb9\u5668\u5316\u73af\u5883\uff0c\u62c9\u53d6\u955c\u50cf\u540e\u542f\u52a8\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>docker pull ollama\/ollama\ndocker run -d -p 11434:11434 --name ollama ollama\/ollama\n<\/code><\/pre>\n\n\n\n<h2 class=\"wp-block-heading\">\u56db\u3001\u6a21\u578b\u7ba1\u7406\u4e0e\u57fa\u7840\u4f7f\u7528<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">1. \u5e38\u7528\u6838\u5fc3\u547d\u4ee4<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u62c9\u53d6\u6a21\u578b\uff1a<code>ollama pull \u6a21\u578b\u540d:\u7248\u672c<\/code>\uff0c\u4f8b\uff1a<code>ollama pull llama3:8b<\/code>\u3001<code>ollama pull qwen2.5:7b<\/code><\/li>\n\n\n\n<li>\u8fd0\u884c\u6a21\u578b\uff1a<code>ollama run \u6a21\u578b\u540d<\/code>\uff0c\u8fdb\u5165\u547d\u4ee4\u884c\u4ea4\u4e92\u6a21\u5f0f<\/li>\n\n\n\n<li>\u67e5\u770b\u672c\u5730\u6a21\u578b\uff1a<code>ollama list<\/code><\/li>\n\n\n\n<li>\u5220\u9664\u6a21\u578b\uff1a<code>ollama rm \u6a21\u578b\u540d<\/code><\/li>\n\n\n\n<li>\u542f\u52a8\u670d\u52a1\uff1a<code>ollama serve<\/code>\uff0c\u9ed8\u8ba4\u76d1\u542c 127.0.0.1:11434<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">2. \u81ea\u5b9a\u4e49\u6a21\u578b\uff08Modelfile\uff09<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">\u901a\u8fc7 Modelfile \u5b9a\u5236\u6a21\u578b\u53c2\u6570\u3001\u7cfb\u7edf\u63d0\u793a\u8bcd\uff0c\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u521b\u5efa Modelfile \u6587\u4ef6\uff0c\u57fa\u7840\u914d\u7f6e\u793a\u4f8b\uff1a<\/li>\n<\/ol>\n\n\n\n<pre class=\"wp-block-code\"><code>FROM qwen2.5:7b\nPARAMETER temperature 0.7\nPARAMETER num_ctx 4096\nSYSTEM \"<em><strong>\u4f60\u7684\u8981\u6c42<\/strong><\/em>\"\n<\/code><\/pre>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u6784\u5efa\u81ea\u5b9a\u4e49\u6a21\u578b\uff1a<code>ollama create custom-model -f Modelfile<\/code><\/li>\n\n\n\n<li>\u8fd0\u884c\u81ea\u5b9a\u4e49\u6a21\u578b\uff1a<code>ollama run custom-model<\/code><\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">3. API \u8c03\u7528<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Ollama \u63d0\u4f9b RESTful API\uff0c\u652f\u6301\u6d41\u5f0f\u4e0e\u975e\u6d41\u5f0f\u5bf9\u8bdd\u8bf7\u6c42\uff0c\u57fa\u7840\u63a5\u53e3\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u7248\u672c\u67e5\u8be2\uff1a<code>http:\/\/localhost:11434\/api\/version<\/code><\/li>\n\n\n\n<li>\u5bf9\u8bdd\u63a5\u53e3\uff1a<code>http:\/\/localhost:11434\/api\/chat<\/code><\/li>\n\n\n\n<li>\u6a21\u578b\u5217\u8868\uff1a<code>http:\/\/localhost:11434\/api\/tags<\/code><\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">4. \u79c1\u6709 API \u914d\u7f6e\uff08\u5c40\u57df\u7f51 \/ \u5185\u7f51\u79c1\u6709\u5316\u90e8\u7f72\uff09<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Ollama \u652f\u6301\u5c06 API \u5f00\u653e\u81f3<strong>\u5c40\u57df\u7f51 \/ \u79c1\u6709\u7f51\u7edc<\/strong>\uff0c\u5e76\u914d\u7f6e\u9274\u6743\u3001\u7aef\u53e3\u3001\u8de8\u57df\uff0c\u5b9e\u73b0\u79c1\u6709\u5316 API \u670d\u52a1\uff1a<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\uff081\uff09\u4fee\u6539 API \u76d1\u542c\u5730\u5740\uff08\u5f00\u653e\u5c40\u57df\u7f51\u8bbf\u95ee\uff09<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">\u9ed8\u8ba4\u4ec5\u672c\u5730\u8bbf\u95ee\uff0c\u4fee\u6539\u73af\u5883\u53d8\u91cf<code>OLLAMA_HOST<\/code>\u5f00\u653e\u79c1\u6709 IP\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Windows\uff08\u7ec8\u7aef\u6267\u884c\uff09\uff1a<\/li>\n<\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>setx OLLAMA_HOST 0.0.0.0:11434\n<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Linux\/macOS\uff08\u7ec8\u7aef\u6267\u884c\uff09\uff1a<\/li>\n<\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>export OLLAMA_HOST=0.0.0.0:11434\n<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Docker \u90e8\u7f72\uff1a<\/li>\n<\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>docker run -d -p 11434:11434 -e OLLAMA_HOST=0.0.0.0:11434 --name ollama ollama\/ollama\n<\/code><\/pre>\n\n\n\n<h4 class=\"wp-block-heading\">\uff082\uff09\u914d\u7f6e API \u5bc6\u94a5\u9274\u6743\uff08\u79c1\u6709 API \u5b89\u5168\u5fc5\u5907\uff09<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">\u8bbe\u7f6e\u73af\u5883\u53d8\u91cf<code>OLLAMA_API_KEY<\/code>\u5f00\u542f\u5bc6\u94a5\u9a8c\u8bc1\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Linux\/macOS\uff1a<\/li>\n<\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>export OLLAMA_API_KEY=\u4f60\u7684\u81ea\u5b9a\u4e49\u5bc6\u94a5\n<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Windows\uff1a<\/li>\n<\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>setx OLLAMA_API_KEY \u4f60\u7684\u81ea\u5b9a\u4e49\u5bc6\u94a5\n<\/code><\/pre>\n\n\n\n<p class=\"wp-block-paragraph\">\u8c03\u7528 API \u65f6\u9700\u5728\u8bf7\u6c42\u5934\u643a\u5e26\u5bc6\u94a5\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>Authorization: Bearer \u4f60\u7684\u81ea\u5b9a\u4e49\u5bc6\u94a5\n<\/code><\/pre>\n\n\n\n<h4 class=\"wp-block-heading\">\uff083\uff09CORS \u8de8\u57df\u914d\u7f6e\uff08\u5bf9\u63a5\u524d\u7aef \/ \u7b2c\u4e09\u65b9\u5e94\u7528\uff09<\/h4>\n\n\n\n<pre class=\"wp-block-code\"><code>export OLLAMA_CORS_ALLOW_ORIGINS=http(s):\/\/\u4f60\u7684\u524d\u7aef\u57df\u540d\n<\/code><\/pre>\n\n\n\n<h4 class=\"wp-block-heading\">\uff084\uff09\u91cd\u542f\u670d\u52a1\u751f\u6548<\/h4>\n\n\n\n<pre class=\"wp-block-code\"><code># \u505c\u6b62\u539f\u6709\u670d\u52a1\npkill ollama\n# \u91cd\u65b0\u542f\u52a8\nollama serve\n<\/code><\/pre>\n\n\n\n<h4 class=\"wp-block-heading\">\uff085\uff09\u79c1\u6709 API \u9a8c\u8bc1<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">\u5c40\u57df\u7f51\u5185\u5176\u4ed6\u8bbe\u5907\u8bbf\u95ee\uff1a<code>http:\/\/\u670d\u52a1\u5668\u5185\u7f51IP:11434\/api\/tags<\/code><\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e94\u3001\u90e8\u7f72\u6838\u5fc3\u6ce8\u610f\u4e8b\u9879<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">1. \u786c\u4ef6\u8d44\u6e90\u7ba1\u63a7<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5185\u5b58\u4e0d\u8db3\u4f1a\u5bfc\u81f4\u6a21\u578b\u5d29\u6e83\u3001\u63a8\u7406\u5361\u987f\uff0c\u4f18\u5148\u9009\u62e9\u4e0e\u786c\u4ef6\u5339\u914d\u7684\u6a21\u578b<\/li>\n\n\n\n<li>\u65e0 GPU \u65f6\u4ec5\u8fd0\u884c 7B \u53ca\u4ee5\u4e0b\u5c0f\u6a21\u578b\uff0c\u907f\u514d CPU \u8d1f\u8f7d\u8fc7\u9ad8<\/li>\n\n\n\n<li>\u542f\u7528 GPU \u52a0\u901f\u9700\u5b89\u88c5\u5bf9\u5e94\u9a71\u52a8\uff0cNVIDIA \u9700\u914d\u7f6e CUDA \u73af\u5883\u53d8\u91cf<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">2. GPU \u663e\u5b58\u4f7f\u7528\u6ce8\u610f\u4e8b\u9879<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4f18\u5148\u9009\u62e9<strong>Q4_K_M \u91cf\u5316\u6a21\u578b<\/strong>\uff0c\u5728\u7cbe\u5ea6\u635f\u5931\u6781\u5c0f\u7684\u524d\u63d0\u4e0b\u8282\u7701 50% \u4ee5\u4e0a\u663e\u5b58<\/li>\n\n\n\n<li>\u663e\u5b58\u63a5\u8fd1\u9608\u503c\u65f6\uff0c\u5173\u95ed\u6d4f\u89c8\u5668\u3001\u6e38\u620f\u7b49\u5360\u7528\u663e\u5b58\u7684\u7a0b\u5e8f<\/li>\n\n\n\n<li>\u591a\u5361\u73af\u5883\u53ef\u901a\u8fc7<code>OLLAMA_NUM_GPU<\/code>\u6307\u5b9a\u663e\u5361\u7f16\u53f7\uff0c\u907f\u514d\u663e\u5b58\u4e89\u62a2<\/li>\n\n\n\n<li>\u663e\u5b58\u4e0d\u8db3\u5f3a\u5236\u52a0\u8f7d\u6a21\u578b\uff0c\u4f1a\u89e6\u53d1\u5185\u5b58\u4ea4\u6362\uff0c\u63a8\u7406\u901f\u5ea6\u4e0b\u964d 90% \u4ee5\u4e0a<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">3. \u7f51\u7edc\u4e0e\u4e0b\u8f7d\u95ee\u9898<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u56fd\u5185\u7f51\u7edc\u4e0b\u8f7d\u6a21\u578b\u901f\u5ea6\u6162\uff0c\u53ef\u5207\u6362\u7a33\u5b9a\u7f51\u7edc\u6216\u4f7f\u7528\u955c\u50cf\u6e90<\/li>\n\n\n\n<li>\u4e0b\u8f7d\u4e2d\u65ad\u65e0\u9700\u91cd\u65b0\u5f00\u59cb\uff0c\u91cd\u65b0\u6267\u884c<code>ollama pull<\/code>\u547d\u4ee4\u652f\u6301\u65ad\u70b9\u7eed\u4f20<\/li>\n\n\n\n<li>\u79bb\u7ebf\u4f7f\u7528\u9700\u63d0\u524d\u5b8c\u6210\u6a21\u578b\u62c9\u53d6\uff0c\u90e8\u7f72\u540e\u53ef\u65ad\u5f00\u7f51\u7edc<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">4. \u5b89\u5168\u9632\u62a4<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u79c1\u6709 API \u5e94\u5728<strong>\u5185\u7f51\u4f7f\u7528<\/strong>\uff0c<strong>\u4e0d\u5efa\u8bae\u516c\u7f51\u76f4\u63a5\u66b4\u9732 11434 \u7aef\u53e3<\/strong><\/li>\n\n\n\n<li>\u516c\u7f51\u8bbf\u95ee\u5e94\u901a\u8fc7 Nginx\/Traefik \u53cd\u5411\u4ee3\u7406 + \u9632\u706b\u5899\u9650\u5236<\/li>\n\n\n\n<li>\u914d\u7f6e API \u5bc6\u94a5\uff0c\u7981\u6b62\u65e0\u9274\u6743\u516c\u5f00\u8c03\u7528<\/li>\n\n\n\n<li>\u4ec5\u4ece Ollama \u5b98\u65b9\u6a21\u578b\u5e93\u62c9\u53d6\u6a21\u578b\uff0c\u62d2\u7edd\u7b2c\u4e09\u65b9\u4e0d\u660e\u6765\u6e90\u6a21\u578b<\/li>\n\n\n\n<li>\u5b9a\u671f\u66f4\u65b0 Ollama\uff0c\u4fee\u590d\u5b89\u5168\u6f0f\u6d1e<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">5. \u5b58\u50a8\u8def\u5f84\u7ba1\u7406<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Windows \u9ed8\u8ba4\u5b58\u50a8\u5728 C \u76d8\uff0c\u6613\u5bfc\u81f4\u7cfb\u7edf\u76d8\u7a7a\u95f4\u4e0d\u8db3\uff0c\u53ef\u4fee\u6539\u5b58\u50a8\u8def\u5f84<\/li>\n\n\n\n<li>Linux\/macOS \u901a\u8fc7\u73af\u5883\u53d8\u91cf<code>OLLAMA_MODELS<\/code>\u6307\u5b9a\u6a21\u578b\u5b58\u50a8\u76ee\u5f55<\/li>\n\n\n\n<li>\u5b9a\u671f\u5220\u9664\u65e0\u7528\u6a21\u578b\uff0c\u91ca\u653e\u5b58\u50a8\u7a7a\u95f4<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">6. \u6743\u9650\u4e0e\u670d\u52a1\u914d\u7f6e<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Windows \u5b89\u88c5\u5fc5\u987b\u7528\u7ba1\u7406\u5458\u6743\u9650\uff0c\u5426\u5219\u73af\u5883\u53d8\u91cf\u914d\u7f6e\u5931\u8d25<\/li>\n\n\n\n<li>Linux \u9700\u786e\u4fdd ollama \u670d\u52a1\u6b63\u5e38\u8fd0\u884c\uff0c\u53ef\u901a\u8fc7<code>systemctl status ollama<\/code>\u67e5\u770b<\/li>\n\n\n\n<li>\u7aef\u53e3\u51b2\u7a81\u65f6\uff0c\u4fee\u6539<code>OLLAMA_HOST<\/code>\u73af\u5883\u53d8\u91cf\u66f4\u6362\u7aef\u53e3<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">7. \u6027\u80fd\u4f18\u5316<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8c03\u6574<code>num_ctx<\/code>\u53c2\u6570\u63a7\u5236\u4e0a\u4e0b\u6587\u7a97\u53e3\u5927\u5c0f\uff0c\u5e73\u8861\u63a8\u7406\u901f\u5ea6\u4e0e\u5185\u5b58\u5360\u7528<\/li>\n\n\n\n<li>GPU \u73af\u5883\u8bbe\u7f6e<code>OLLAMA_NUM_GPU<\/code>\u5206\u914d\u663e\u5b58\u6bd4\u4f8b<\/li>\n\n\n\n<li>\u5173\u95ed\u540e\u53f0\u5360\u7528\u8d44\u6e90\u7684\u7a0b\u5e8f\uff0c\u63d0\u5347\u63a8\u7406\u54cd\u5e94\u901f\u5ea6<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">\u516d\u3001\u5e38\u89c1\u95ee\u9898\u6392\u67e5<\/h2>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u5b89\u88c5\u540e\u547d\u4ee4\u65e0\u6548\uff1aWindows \u91cd\u542f\u7ec8\u7aef\uff0cLinux\/macOS \u91cd\u65b0\u52a0\u8f7d\u73af\u5883\u53d8\u91cf<\/li>\n\n\n\n<li>\u6a21\u578b\u542f\u52a8\u5931\u8d25\uff1a\u68c0\u67e5\u5185\u5b58 \/ \u663e\u5b58\u662f\u5426\u5145\u8db3\uff0c\u66f4\u6362\u5c0f\u91cf\u5316\u6a21\u578b\u5c1d\u8bd5<\/li>\n\n\n\n<li>\u7aef\u53e3\u88ab\u5360\u7528\uff1a\u4f7f\u7528<code>netstat<\/code>\u547d\u4ee4\u67e5\u770b\u7aef\u53e3\u5360\u7528\uff0c\u66f4\u6362\u76d1\u542c\u7aef\u53e3<\/li>\n\n\n\n<li>\u4e0b\u8f7d\u901f\u5ea6\u4e3a 0\uff1a\u5207\u6362\u7f51\u7edc\u3001\u5173\u95ed\u4ee3\u7406\uff0c\u91cd\u65b0\u6267\u884c\u62c9\u53d6\u547d\u4ee4<\/li>\n\n\n\n<li>GPU \u672a\u542f\u7528\uff1a\u68c0\u67e5\u9a71\u52a8\u7248\u672c\uff0c\u786e\u8ba4\u663e\u5361\u652f\u6301\u5bf9\u5e94\u52a0\u901f\u6846\u67b6<\/li>\n\n\n\n<li>\u79c1\u6709 API \u65e0\u6cd5\u8bbf\u95ee\uff1a\u68c0\u67e5\u9632\u706b\u5899\u7aef\u53e3\u653e\u884c\u3001\u76d1\u542c\u5730\u5740\u914d\u7f6e\u3001API \u5bc6\u94a5\u662f\u5426\u6b63\u786e<\/li>\n\n\n\n<li>\u63a8\u7406\u6781\u6162\uff1a\u663e\u5b58\u4e0d\u8db3\u89e6\u53d1 CPU \u4ea4\u6362\uff0c\u964d\u4f4e\u6a21\u578b\u53c2\u6570\u91cf\u6216\u66f4\u6362\u91cf\u5316\u7248\u672c<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"Ollama\u662f\u4e00\u6b3e\u5f00\u6e90\u8de8\u5e73\u53f0\u7684\u5927\u6a21\u578b\u8fd0\u884c\u5de5\u5177\uff0c\u652f\u6301\u4e00\u952e\u90e8\u7f72Llama\u3001Qwen\u7b49\u4e3b\u6d41\u6a21\u578b\uff0c\u4fdd\u969c\u6570\u636e\u9690\u79c1\u4e0e\u79bb\u7ebf\u4f7f\u7528\u3002\u5b89\u88c5\u7b80\u5355\uff0c\u652f\u6301Windows\u3001macOS\u3001Linux\u53caDocker\u3002\u90e8\u7f72\u9700\u5173\u6ce8\u786c\u4ef6\u914d\u7f6e\uff0c\u7279\u522b\u662fGPU\u663e\u5b58\u8981","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"emotion":"","emotion_color":"","title_style":"","license":"","footnotes":""},"categories":[5],"tags":[],"class_list":["post-61","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/posts\/61","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/comments?post=61"}],"version-history":[{"count":2,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/posts\/61\/revisions"}],"predecessor-version":[{"id":63,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/posts\/61\/revisions\/63"}],"wp:attachment":[{"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/media?parent=61"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/categories?post=61"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.dinonode.org\/index.php\/wp-json\/wp\/v2\/tags?post=61"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}