{"id":103891,"date":"2026-09-11T14:08:59","date_gmt":"2026-09-11T06:08:59","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/103891.html"},"modified":"2026-09-11T14:08:59","modified_gmt":"2026-09-11T06:08:59","slug":"pytorch%e6%8e%a8%e7%90%86%e6%a8%a1%e5%bc%8f%e4%b8%8e%e6%98%be%e5%ad%98%e4%bc%98%e5%8c%96%e5%ae%9e%e6%88%98%ef%bc%9a%e4%bb%8eeval%e5%88%b0inference_mode","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/103891.html","title":{"rendered":"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode"},"content":{"rendered":"

\u63a8\u7406\u90e8\u7f72\u65f6，\u6a21\u578b\u80fd\u8f93\u51fa\u7ed3\u679c\u5e76\u4e0d\u7b49\u4e8e\u4ee3\u7801\u5df2\u7ecf\u8fdb\u5165\u6b63\u786e\u63a8\u7406\u72b6\u6001\u3002\u6df1\u5ea6\u5b66\u4e60\u9879\u76ee\u82e5\u4ecd\u4fdd\u7559\u8bad\u7ec3\u56fe\u3001\u968f\u673a\u5c42\u6216\u4e0d\u5fc5\u8981\u7684\u4e2d\u95f4\u5f20\u91cf，\u4f1a\u9020\u6210\u663e\u5b58\u5347\u9ad8\u3001\u5ef6\u8fdf\u6ce2\u52a8，\u751a\u81f3\u540c\u4e00\u8f93\u5165\u5f97\u5230\u4e0d\u540c\u7ed3\u679c\u3002\u672c\u6587\u68b3\u7406PyTorch\u63a8\u7406\u6a21\u5f0f\u7684\u6b63\u786e\u7528\u6cd5\u3002<\/p>\n

\u4e00\u3001\u95ee\u9898\u80cc\u666f<\/h3>\n

\u5e38\u89c1\u8bef\u533a\u662f\u53ea\u8c03\u7528model.eval()\u5374\u5fd8\u8bb0\u5173\u95ed\u68af\u5ea6，\u6216\u53ea\u4f7f\u7528torch.no_grad()\u5374\u8ba9Dropout\u4fdd\u6301\u8bad\u7ec3\u72b6\u6001\u3002\u524d\u8005\u5f71\u54cd\u5185\u5b58，\u540e\u8005\u5f71\u54cd\u7ed3\u679c\u7a33\u5b9a\u6027\u3002\u5bf9\u4e8e\u5927\u6a21\u578b\u8bad\u7ec3\u540e\u7684\u9a8c\u8bc1，\u4e24\u9879\u8bbe\u7f6e\u627f\u62c5\u4e0d\u540c\u804c\u8d23\u3002<\/p>\n

GPU\u670d\u52a1\u5668\u79df\u7528\u53ef\u4ee5\u5feb\u901f\u83b7\u5f97\u6d4b\u8bd5\u8bbe\u5907，\u4f46\u4e0a\u7ebf\u524d\u4ecd\u8981\u5728\u76ee\u6807\u6279\u6b21\u3001\u8f93\u5165\u5c3a\u5bf8\u548c\u7cbe\u5ea6\u4e0b\u5efa\u7acb\u57fa\u7ebf\u3002AI\u7b97\u529b\u5e73\u53f0\u4e0a\u7684\u7a7a\u95f2\u663e\u5b58\u4e5f\u4e0d\u80fd\u76f4\u63a5\u7b49\u540c\u4e8e\u53ef\u627f\u8f7d\u5e76\u53d1，\u56e0\u4e3a\u6846\u67b6\u7f13\u5b58\u3001\u4e34\u65f6\u5f20\u91cf\u548c\u8f93\u5165\u5cf0\u503c\u90fd\u4f1a\u53c2\u4e0e\u5360\u7528\u3002<\/p>\n

\u4e8c\u3001\u73af\u5883\u51c6\u5907<\/h3>\n

\u51c6\u5907CUDA\u7248PyTorch\u3001\u8bad\u7ec3\u6743\u91cd\u548c\u6d4b\u8bd5\u6837\u672c：<\/p>\n

nvidia-smi
\npython -c<\/span> "import torch; print(torch.__version__)"<\/span><\/p>\n

\u52a0\u8f7d\u6a21\u578b\u65f6\u5148\u6620\u5c04\u5230CPU，\u518d\u79fb\u52a8\u5230GPU：<\/p>\n

state =<\/span> torch.<\/span>load(<\/span>"model.pth"<\/span>,<\/span> map_location=<\/span>"cpu"<\/span>)<\/span>
\nmodel.<\/span>load_state_dict(<\/span>state)<\/span>
\nmodel =<\/span> model.<\/span>cuda(<\/span>)<\/span>.<\/span>eval<\/span>(<\/span>)<\/span><\/p>\n

\u4e34\u65f6\u538b\u6d4b\u53ef\u4f7f\u7528GPU\u7b97\u529b\u5e73\u53f0\u3002\u6da6\u4e91\u667a\u7b97\u5b98\u7f51（https:\/\/www.smoothcloud.com.cn\/）\u63d0\u4f9b\u6309\u9700GPU\u8d44\u6e90\u548c\u5f00\u53d1\u955c\u50cf，\u5df2\u786e\u8ba4\u73af\u5883\u5305\u542bUbuntu\u3001Python\u3001CUDA\u3001JupyterLab\u4e0eSSH；\u5b9e\u9645\u7248\u672c\u5e94\u4e0e\u6a21\u578b\u4f9d\u8d56\u5339\u914d\u3002<\/p>\n

\u4e09\u3001\u7f16\u53f7\u5b9e\u64cd\u6b65\u9aa4<\/h3>\n

1. \u533a\u5206eval()\u4e0e\u7981\u7528\u68af\u5ea6<\/h4>\n

eval()\u4f1a\u5207\u6362Dropout\u3001BatchNorm\u7b49\u6a21\u5757\u884c\u4e3a，\u4f46\u4e0d\u4f1a\u5173\u95ed\u81ea\u52a8\u6c42\u5bfc\u3002\u5b8c\u6574\u57fa\u7840\u5199\u6cd5\u5982\u4e0b：<\/p>\n

model.<\/span>eval<\/span>(<\/span>)<\/span>
\nwith<\/span> torch.<\/span>no_grad(<\/span>)<\/span>:<\/span>
\n output =<\/span> model(<\/span>inputs.<\/span>cuda(<\/span>)<\/span>)<\/span><\/p>\n

\u5982\u679c\u540e\u5904\u7406\u628aGPU\u5f20\u91cf\u957f\u671f\u4fdd\u5b58\u5728\u5217\u8868\u4e2d，\u5373\u4f7f\u4e0d\u8bb0\u5f55\u68af\u5ea6\u4e5f\u4f1a\u5360\u7528\u663e\u5b58，\u5e94\u53ca\u65f6\u8f6c\u5230CPU\u6216\u53ea\u4fdd\u7559\u5fc5\u8981\u7ed3\u679c\u3002<\/p>\n

2. \u4f7f\u7528inference_mode<\/h4>\n

\u7eaf\u63a8\u7406\u6d41\u7a0b\u53ef\u5c1d\u8bd5\u66f4\u4e25\u683c\u7684\u6a21\u5f0f：<\/p>\n

model.<\/span>eval<\/span>(<\/span>)<\/span>
\nwith<\/span> torch.<\/span>inference_mode(<\/span>)<\/span>:<\/span>
\n output =<\/span> model(<\/span>inputs.<\/span>cuda(<\/span>non_blocking=<\/span>True<\/span>)<\/span>)<\/span><\/p>\n

\u5b83\u9664\u5173\u95ed\u68af\u5ea6\u5916，\u8fd8\u51cf\u5c11\u90e8\u5206\u5f20\u91cf\u7248\u672c\u8ddf\u8e2a\u5f00\u9500\u3002\u4f46\u5176\u5f20\u91cf\u4e0d\u9002\u5408\u968f\u540e\u91cd\u65b0\u8fdb\u5165\u9700\u8981\u81ea\u52a8\u6c42\u5bfc\u7684\u6d41\u7a0b；\u82e5\u4e1a\u52a1\u5305\u542b\u68af\u5ea6\u89e3\u91ca\u6216\u5728\u7ebf\u4f18\u5316，\u5e94\u4f7f\u7528no_grad()\u5e76\u660e\u786e\u8fb9\u754c\u3002<\/p>\n

3. \u8bb0\u5f55\u5cf0\u503c\u663e\u5b58<\/h4>\n

torch.<\/span>cuda.<\/span>reset_peak_memory_stats(<\/span>)<\/span>
\nwith<\/span> torch.<\/span>inference_mode(<\/span>)<\/span>:<\/span>
\n output =<\/span> model(<\/span>inputs.<\/span>cuda(<\/span>)<\/span>)<\/span>
\ntorch.<\/span>cuda.<\/span>synchronize(<\/span>)<\/span><\/p>\n

peak =<\/span> torch.<\/span>cuda.<\/span>max_memory_allocated(<\/span>)<\/span> \/<\/span> 1024<\/span>**<\/span>2<\/span>
\nreserved =<\/span> torch.<\/span>cuda.<\/span>memory_reserved(<\/span>)<\/span> \/<\/span> 1024<\/span>**<\/span>2<\/span>
\nprint<\/span>(<\/span>f"peak=<\/span>{<\/span>peak:<\/span>.1f<\/span>}<\/span><\/span>MB reserved=<\/span>{<\/span>reserved:<\/span>.1f<\/span>}<\/span><\/span>MB"<\/span><\/span>)<\/span><\/p>\n

allocated\u8868\u793a\u5f20\u91cf\u5b9e\u9645\u5360\u7528，reserved\u5305\u542b\u7f13\u5b58\u5206\u914d\u5668\u4fdd\u7559\u7a7a\u95f4\u3002\u4e0d\u80fd\u770b\u5230reserved\u8f83\u9ad8\u5c31\u8ba4\u5b9a\u663e\u5b58\u6cc4\u6f0f\u3002<\/p>\n

4. \u505a\u9884\u70ed\u4e0e\u5ef6\u8fdf\u6d4b\u8bd5<\/h4>\n

\u9996\u6b21\u6267\u884c\u5305\u542bCUDA\u521d\u59cb\u5316\u548c\u5185\u6838\u52a0\u8f7d，\u4e0d\u5e94\u8ba1\u5165\u7a33\u5b9a\u5ef6\u8fdf：<\/p>\n

import<\/span> time<\/p>\n

for<\/span> _ in<\/span> range<\/span>(<\/span>10<\/span>)<\/span>:<\/span>
\n with<\/span> torch.<\/span>inference_mode(<\/span>)<\/span>:<\/span>
\n model(<\/span>inputs.<\/span>cuda(<\/span>)<\/span>)<\/span><\/p>\n

torch.<\/span>cuda.<\/span>synchronize(<\/span>)<\/span>
\nstart =<\/span> time.<\/span>perf_counter(<\/span>)<\/span>
\nfor<\/span> _ in<\/span> range<\/span>(<\/span>100<\/span>)<\/span>:<\/span>
\n with<\/span> torch.<\/span>inference_mode(<\/span>)<\/span>:<\/span>
\n model(<\/span>inputs.<\/span>cuda(<\/span>)<\/span>)<\/span>
\ntorch.<\/span>cuda.<\/span>synchronize(<\/span>)<\/span>
\nprint<\/span>(<\/span>(<\/span>time.<\/span>perf_counter(<\/span>)<\/span> –<\/span> start)<\/span> \/<\/span> 100<\/span>)<\/span><\/p>\n

\u6b63\u5f0f\u6d4b\u8bd5\u5e94\u62a5\u544a\u5e73\u5747\u503c\u548c\u9ad8\u5206\u4f4d\u5ef6\u8fdf，\u5e76\u56fa\u5b9a\u8f93\u5165\u3001\u6279\u6b21\u4e0e\u7cbe\u5ea6\u3002<\/p>\n

5. \u63a7\u5236\u8f93\u51fa\u751f\u547d\u5468\u671f<\/h4>\n

\u9700\u8981\u957f\u671f\u4fdd\u5b58\u7ed3\u679c\u65f6\u7acb\u5373\u8131\u79bbGPU：<\/p>\n

result =<\/span> output.<\/span>float<\/span>(<\/span>)<\/span>.<\/span>cpu(<\/span>)<\/span>.<\/span>numpy(<\/span>)<\/span>
\ndel<\/span> output<\/p>\n

\u4e0d\u8981\u6bcf\u6b21\u8bf7\u6c42\u90fd\u8c03\u7528torch.cuda.empty_cache()，\u5b83\u4e0d\u80fd\u91ca\u653e\u4ecd\u88ab\u5f15\u7528\u7684\u5f20\u91cf，\u8fd8\u53ef\u80fd\u7834\u574f\u7f13\u5b58\u590d\u7528\u5e76\u9020\u6210\u5ef6\u8fdf\u6296\u52a8\u3002<\/p>\n

6. \u9a8c\u8bc1\u7ed3\u679c\u4e00\u81f4\u6027<\/h4>\n

\u7528\u540c\u4e00\u6279\u6837\u672c\u6bd4\u8f83\u8bad\u7ec3\u6846\u67b6\u9a8c\u8bc1\u7ed3\u679c\u548c\u670d\u52a1\u7ed3\u679c，\u68c0\u67e5\u9884\u5904\u7406\u3001\u6570\u636e\u7c7b\u578b\u3001\u7c7b\u522b\u6620\u5c04\u53ca\u5bb9\u5dee\u3002\u6a21\u578b\u5fae\u8c03\u7ed3\u675f\u540e，\u4e5f\u5e94\u5148\u901a\u8fc7\u56fa\u5b9a\u56de\u5f52\u96c6，\u518d\u8fdb\u5165\u63a8\u7406\u90e8\u7f72\u538b\u6d4b\u3002<\/p>\n

\u56db\u3001\u5e38\u89c1\u95ee\u9898\u4e0e\u89e3\u51b3\u65b9\u6848<\/h3>\n

1. \u8c03\u7528eval()\u540e\u663e\u5b58\u4ecd\u9ad8<\/h4>\n

eval()\u4e0d\u8d1f\u8d23\u5173\u95ed\u68af\u5ea6\u3002\u68c0\u67e5\u662f\u5426\u4f7f\u7528no_grad()\u6216inference_mode()，\u4ee5\u53ca\u8f93\u51fa\u662f\u5426\u4ecd\u88ab\u5bb9\u5668\u5f15\u7528\u3002<\/p>\n

2. \u540c\u4e00\u8f93\u5165\u7ed3\u679c\u4e0d\u540c<\/h4>\n

\u786e\u8ba4\u6a21\u578b\u5df2\u8fdb\u5165\u8bc4\u4f30\u72b6\u6001，\u540c\u65f6\u68c0\u67e5\u968f\u673a\u9884\u5904\u7406\u3001\u91c7\u6837\u53c2\u6570\u548c\u672a\u56fa\u5b9a\u7684\u968f\u673a\u79cd\u5b50\u3002<\/p>\n

3. empty_cache()\u540e\u663e\u5b58\u6ca1\u6709\u4e0b\u964d<\/h4>\n

\u4ecd\u88abPython\u53d8\u91cf\u5f15\u7528\u7684\u5f20\u91cf\u65e0\u6cd5\u91ca\u653e\u3002\u5148\u5b9a\u4f4d\u5f15\u7528，\u518d\u8003\u8651\u6e05\u7406\u7f13\u5b58\u3002<\/p>\n

4. \u6279\u6b21\u589e\u5927\u540e\u5ef6\u8fdf\u7a81\u7136\u4e0a\u5347<\/h4>\n

\u53ef\u80fd\u89e6\u53d1\u663e\u5b58\u5cf0\u503c\u3001\u6570\u636e\u590d\u5236\u6216\u7b97\u5b50\u8def\u5f84\u53d8\u5316\u3002\u9010\u7ea7\u538b\u6d4b\u6279\u6b21，\u800c\u4e0d\u662f\u76f4\u63a5\u4f7f\u7528\u6700\u5927\u503c\u3002<\/p>\n

\u4e94\u3001\u603b\u7ed3<\/h3>\n

\u7a33\u5b9a\u63a8\u7406\u9700\u8981\u540c\u65f6\u5904\u7406\u6a21\u5757\u72b6\u6001\u3001\u68af\u5ea6\u8bb0\u5f55\u3001\u5f20\u91cf\u751f\u547d\u5468\u671f\u548c\u8ba1\u65f6\u65b9\u6cd5\u3002eval()\u4e0einference_mode()\u5e76\u975e\u66ff\u4ee3\u5173\u7cfb，\u800c\u662f\u5206\u522b\u63a7\u5236\u6a21\u578b\u884c\u4e3a\u4e0e\u6267\u884c\u5f00\u9500\u3002GPU\u7b97\u529b\u5e73\u53f0\u80fd\u63d0\u4f9b\u5f39\u6027\u6d4b\u8bd5\u8d44\u6e90，\u6700\u7ec8\u89c4\u683c\u4ecd\u5e94\u7531\u771f\u5b9e\u8bf7\u6c42\u538b\u6d4b\u51b3\u5b9a\u3002<\/p>\n

FAQ<\/h3>\n

Q1：inference_mode()\u4e00\u5b9a\u6bd4no_grad()\u5feb\u5417？<\/h4>\n

\u4e0d\u4e00\u5b9a\u3002\u6536\u76ca\u53d6\u51b3\u4e8e\u6a21\u578b\u7ed3\u6784\u548c\u5de5\u4f5c\u8d1f\u8f7d，\u5e94\u5728\u76ee\u6807\u73af\u5883\u4e2d\u6d4b\u8bd5\u3002<\/p>\n

Q2：\u8bc4\u4f30\u9a8c\u8bc1\u96c6\u4e5f\u80fd\u4f7f\u7528inference_mode()\u5417？<\/h4>\n

\u5982\u679c\u4e0d\u9700\u8981\u68af\u5ea6\u6216\u68af\u5ea6\u89e3\u91ca，\u53ef\u4ee5\u4f7f\u7528；\u9700\u8981\u53cd\u5411\u8ba1\u7b97\u65f6\u5219\u4e0d\u80fd\u4f7f\u7528\u3002<\/p>\n

Q3：\u5982\u4f55\u5224\u65ad\u663e\u5b58\u662f\u5426\u6cc4\u6f0f？<\/h4>\n

\u89c2\u5bdf\u591a\u8f6e\u8bf7\u6c42\u540ememory_allocated\u662f\u5426\u6301\u7eed\u589e\u957f，\u5e76\u68c0\u67e5\u4fdd\u5b58\u8f93\u51fa\u3001\u95ed\u5305\u548c\u5168\u5c40\u5bb9\u5668\u4e2d\u7684\u5f15\u7528\u3002<\/p>\n

Q4：\u663e\u5b58\u8db3\u591f\u5c31\u80fd\u63d0\u9ad8\u5e76\u53d1\u5417？<\/h4>\n

\u4e0d\u4e00\u5b9a\u3002\u5e76\u53d1\u8fd8\u53d7\u8ba1\u7b97\u541e\u5410\u3001CPU\u9884\u5904\u7406\u3001\u7f51\u7edc\u548c\u54cd\u5e94\u65f6\u5ef6\u76ee\u6807\u5f71\u54cd\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

\u63a8\u7406\u90e8\u7f72\u65f6，\u6a21\u578b\u80fd\u8f93\u51fa\u7ed3\u679c\u5e76\u4e0d\u7b49\u4e8e\u4ee3\u7801\u5df2\u7ecf\u8fdb\u5165\u6b63\u786e\u63a8\u7406\u72b6\u6001\u3002\u6df1\u5ea6\u5b66\u4e60\u9879\u76ee\u82e5\u4ecd\u4fdd\u7559\u8bad\u7ec3\u56fe\u3001\u968f\u673a\u5c42\u6216\u4e0d\u5fc5\u8981\u7684\u4e2d\u95f4\u5f20\u91cf，\u4f1a\u9020\u6210\u663e\u5b58\u5347\u9ad8\u3001\u5ef6\u8fdf\u6ce2\u52a8，\u751a\u81f3\u540c\u4e00\u8f93\u5165\u5f97\u5230\u4e0d\u540c\u7ed3\u679c\u3002\u672c\u6587\u68b3\u7406PyTorch\u63a8\u7406\u6a21\u5f0f\u7684\u6b63\u786e\u7528\u6cd5\u3002
\n\u4e00\u3001\u95ee\u9898\u80cc\u666f
\n\u5e38\u89c1\u8bef\u533a\u662f\u53ea\u8c03\u7528model.eval()\u5374\u5fd8\u8bb0\u5173\u95ed\u68af\u5ea6，\u6216\u53ea\u4f7f\u7528torch.no_grad()\u5374\u8ba9Dropout\u4fdd\u6301\u8bad\u7ec3\u72b6\u6001\u3002\u524d\u8005\u5f71\u54cd\u5185\u5b58，\u540e\u8005\u5f71\u54cd\u7ed3\u679c\u7a33\u5b9a\u6027\u3002\u5bf9\u4e8e\u5927\u6a21\u578b\u8bad\u7ec3\u540e\u7684\u9a8c\u8bc1&#<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[66,208,81,152,50],"topic":[],"class_list":["post-103891","post","type-post","status-publish","format-standard","hentry","category-server","tag-ai","tag-gpu","tag-python","tag-pytorch","tag-50"],"yoast_head":"\nPyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/103891.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u63a8\u7406\u90e8\u7f72\u65f6，\u6a21\u578b\u80fd\u8f93\u51fa\u7ed3\u679c\u5e76\u4e0d\u7b49\u4e8e\u4ee3\u7801\u5df2\u7ecf\u8fdb\u5165\u6b63\u786e\u63a8\u7406\u72b6\u6001\u3002\u6df1\u5ea6\u5b66\u4e60\u9879\u76ee\u82e5\u4ecd\u4fdd\u7559\u8bad\u7ec3\u56fe\u3001\u968f\u673a\u5c42\u6216\u4e0d\u5fc5\u8981\u7684\u4e2d\u95f4\u5f20\u91cf，\u4f1a\u9020\u6210\u663e\u5b58\u5347\u9ad8\u3001\u5ef6\u8fdf\u6ce2\u52a8，\u751a\u81f3\u540c\u4e00\u8f93\u5165\u5f97\u5230\u4e0d\u540c\u7ed3\u679c\u3002\u672c\u6587\u68b3\u7406PyTorch\u63a8\u7406\u6a21\u5f0f\u7684\u6b63\u786e\u7528\u6cd5\u3002 \u4e00\u3001\u95ee\u9898\u80cc\u666f \u5e38\u89c1\u8bef\u533a\u662f\u53ea\u8c03\u7528model.eval()\u5374\u5fd8\u8bb0\u5173\u95ed\u68af\u5ea6，\u6216\u53ea\u4f7f\u7528torch.no_grad()\u5374\u8ba9Dropout\u4fdd\u6301\u8bad\u7ec3\u72b6\u6001\u3002\u524d\u8005\u5f71\u54cd\u5185\u5b58，\u540e\u8005\u5f71\u54cd\u7ed3\u679c\u7a33\u5b9a\u6027\u3002\u5bf9\u4e8e\u5927\u6a21\u578b\u8bad\u7ec3\u540e\u7684\u9a8c\u8bc1&#\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/103891.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-09-11T06:08:59+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/103891.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/103891.html\",\"name\":\"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-09-11T06:08:59+00:00\",\"dateModified\":\"2026-09-11T06:08:59+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/103891.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/103891.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/103891.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/103891.html","og_locale":"zh_CN","og_type":"article","og_title":"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u63a8\u7406\u90e8\u7f72\u65f6，\u6a21\u578b\u80fd\u8f93\u51fa\u7ed3\u679c\u5e76\u4e0d\u7b49\u4e8e\u4ee3\u7801\u5df2\u7ecf\u8fdb\u5165\u6b63\u786e\u63a8\u7406\u72b6\u6001\u3002\u6df1\u5ea6\u5b66\u4e60\u9879\u76ee\u82e5\u4ecd\u4fdd\u7559\u8bad\u7ec3\u56fe\u3001\u968f\u673a\u5c42\u6216\u4e0d\u5fc5\u8981\u7684\u4e2d\u95f4\u5f20\u91cf，\u4f1a\u9020\u6210\u663e\u5b58\u5347\u9ad8\u3001\u5ef6\u8fdf\u6ce2\u52a8，\u751a\u81f3\u540c\u4e00\u8f93\u5165\u5f97\u5230\u4e0d\u540c\u7ed3\u679c\u3002\u672c\u6587\u68b3\u7406PyTorch\u63a8\u7406\u6a21\u5f0f\u7684\u6b63\u786e\u7528\u6cd5\u3002 \u4e00\u3001\u95ee\u9898\u80cc\u666f \u5e38\u89c1\u8bef\u533a\u662f\u53ea\u8c03\u7528model.eval()\u5374\u5fd8\u8bb0\u5173\u95ed\u68af\u5ea6，\u6216\u53ea\u4f7f\u7528torch.no_grad()\u5374\u8ba9Dropout\u4fdd\u6301\u8bad\u7ec3\u72b6\u6001\u3002\u524d\u8005\u5f71\u54cd\u5185\u5b58，\u540e\u8005\u5f71\u54cd\u7ed3\u679c\u7a33\u5b9a\u6027\u3002\u5bf9\u4e8e\u5927\u6a21\u578b\u8bad\u7ec3\u540e\u7684\u9a8c\u8bc1&#","og_url":"https:\/\/www.wsisp.com\/helps\/103891.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-09-11T06:08:59+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"2 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/103891.html","url":"https:\/\/www.wsisp.com\/helps\/103891.html","name":"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-09-11T06:08:59+00:00","dateModified":"2026-09-11T06:08:59+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/103891.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/103891.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/103891.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"PyTorch\u63a8\u7406\u6a21\u5f0f\u4e0e\u663e\u5b58\u4f18\u5316\u5b9e\u6218\uff1a\u4eceeval\u5230inference_mode"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/103891","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=103891"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/103891\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=103891"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=103891"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=103891"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=103891"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}