/tags/%E6%95%85%E9%9A%9C%E6%8E%92%E9%99%A4/Recent content in 故障排除 on Mi&Bee BlogHugo -- gohugo.iozh-CN蓝宝石的傻话Sat, 23 May 2026 00:00:00 +0000/posts/aihelper/yolo-faq/Sat, 23 May 2026 00:00:00 +0000/posts/aihelper/yolo-faq/<h2 id="环境安装问题">环境安装问题</h2> <p><strong>Q1: CUDA 不可用，只使用 CPU？</strong></p> <p>首先确认 NVIDIA 驱动版本是否支持所需 CUDA 版本，驱动版本过低会导致 CUDA 不可用：</p> <div class="code-block-wrapper" data-lang="bash"> <div class="code-block-header"> <div class="code-block-meta"><span class="code-language">bash</span></div> <button class="copy-button" aria-label="Copy code"> <svg class="copy-icon" xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><rect x="9" y="9" width="13" height="13" rx="2" ry="2"/><path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"/></svg> <svg class="check-icon" xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="20 6 9 17 4 12"/></svg> </button> </div> <div class="code-block-body"><div class="highlight"><div class="chroma"> <table class="lntable"><tr><td class="lntd"> <pre tabindex="0" class="chroma"><code><span class="lnt">1 </span><span class="lnt">2 </span><span class="lnt">3 </span><span class="lnt">4 </span><span class="lnt">5 </span><span class="lnt">6 </span></code></pre></td> <td class="lntd"> <pre tabindex="0" class="chroma"><code class="language-bash" data-lang="bash"><span class="line"><span class="cl"><span class="c1"># 检查驱动版本（Driver Version 需 &gt;= 支持CUDA的最低版本）</span> </span></span><span class="line"><span class="cl">nvidia-smi </span></span><span class="line"><span class="cl"><span class="c1"># 检查CUDA工具包版本</span> </span></span><span class="line"><span class="cl">nvcc --version </span></span><span class="line"><span class="cl"><span class="c1"># 重新安装对应CUDA版本的PyTorch</span> </span></span><span class="line"><span class="cl">pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu121</span></span></code></pre></td></tr></table> </div> </div></div> </div><p>如果 <code>nvidia-smi</code> 显示 CUDA 版本但 PyTorch 仍使用 CPU，说明安装的是 CPU 版 PyTorch。卸载后用 <code>--index-url</code> 重新安装 CUDA 版本。CUDA 11.8 用户将 URL 中的 <code>cu121</code> 替换为 <code>cu118</code>。建议使用 conda 或 venv 虚拟环境隔离不同 CUDA 版本的 PyTorch，避免系统级冲突。</p>