[{"data":1,"prerenderedAt":774},["ShallowReactive",2],{"\u002Fposts\u002F37561f92":3,"surround-\u002Fposts\u002F37561f92":762},{"id":4,"title":5,"body":6,"categories":736,"date":738,"description":739,"draft":740,"extension":741,"image":742,"meta":743,"navigation":745,"path":746,"permalink":746,"published":747,"readingTime":748,"recommend":753,"references":747,"seo":754,"sitemap":755,"stem":756,"tags":757,"type":760,"updated":738,"__hash__":761},"content\u002Fposts\u002F2020\u002F利用PyCharm简化搭建深度学习环境(新手排坑).md","利用PyCharm简化搭建深度学习环境(新手排坑)",{"type":7,"value":8,"toc":720},"minimark",[9,37,42,55,72,76,131,209,214,221,230,235,239,257,269,274,279,283,288,292,310,314,319,341,345,374,378,396,525,539,543,548,596,604,611,716],[10,11,12],"blockquote",{},[13,14,15,19,22,25,28,31,34],"ul",{},[16,17,18],"li",{},"预备环境：Anaconda3",[16,20,21],{},"搭建环境 ：PyCharm + TensorFlow\u002FTensorFlow-GPU + Keras",[16,23,24],{},"PyCharm Version 2020.2.1",[16,26,27],{},"TensorFlow Version 2.3.1",[16,29,30],{},"Keras Version 2.4.3",[16,32,33],{},"CUDA Version 10.1",[16,35,36],{},"cuDNN Version 7.6",[38,39,41],"h2",{"id":40},"_1-用-pycharm-新建-python-环境","1. 用 PyCharm 新建 Python 环境",[10,43,44],{},[45,46,47,48],"p",{},"需要提前安装好 Anaconda 官网下载地址：",[49,50,54],"a",{"href":51,"rel":52},"https:\u002F\u002Fwww.anaconda.com\u002Fproducts\u002Findividual",[53],"nofollow","Anaconda 官网下载地址",[56,57,58,66],"ol",{},[16,59,60,61],{},"点击新建项目(New Project)",[62,63],"img",{"alt":64,"src":65},"","https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fnbsnufji.png",[16,67,68,69],{},"按下图操作",[62,70],{"alt":64,"src":71},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fblpxfujj.png",[38,73,75],{"id":74},"_2-安装-tensorflow-231","2. 安装 TensorFlow 2.3.1",[56,77,78,84,102,119,128],{},[16,79,80,81],{},"项目搭建好后点击 Terminal 打开终端",[62,82],{"alt":64,"src":83},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Frwwfbntr.png",[16,85,86,87,91,92,95,96,99],{},"输入命令 ",[88,89,90],"code",{"code":90},"pip install --upgrade pip -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\u002F"," ，从清华的镜像源检查 pip，确保 pip 为最新版，目前 ",[88,93,94],{"code":94},"pip"," 最新版为 ",[88,97,98],{"code":98},"20.2.3",[62,100],{"alt":64,"src":101},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Foxhxunrm.png",[16,103,104,105,108,109,112,113,115,116,118],{},"输入",[88,106,107],{"code":107},"TensorFlow","安装命令 ",[88,110,111],{"code":111},"pip install tensorflow -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\u002F"," 安装最新版的 ",[88,114,107],{"code":107},"，目前最新版的 ",[88,117,107],{"code":107}," 为 2.3.1",[16,120,121,122,125],{},"如果上面安装不了或者下载缓慢，可以试试这条命令，切换为阿里镜像源 ",[88,123,124],{"code":124},"pip install tensorflow -i https:\u002F\u002Fmirrors.aliyun.com\u002Fpypi\u002Fsimple",[62,126],{"alt":64,"src":127},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fthvalpvn.png",[16,129,130],{},"输入以下 Python 代码运行，验证安装是否成功：",[132,133,137],"pre",{"className":134,"code":135,"language":136,"meta":64,"style":64},"language-python shiki shiki-themes catppuccin-latte one-dark-pro","import tensorflow as tf\nprint(tf.reduce_sum(tf.random.normal([1000, 1000])))\n","python",[88,138,139,158],{"__ignoreMap":64},[140,141,144,148,152,155],"span",{"class":142,"line":143},"line",1,[140,145,147],{"class":146},"sSWcl","import",[140,149,151],{"class":150},"sa2x1"," tensorflow ",[140,153,154],{"class":146},"as",[140,156,157],{"class":150}," tf\n",[140,159,161,165,169,172,175,179,181,183,185,188,190,193,196,200,203,206],{"class":142,"line":160},2,[140,162,164],{"class":163},"sk-YW","print",[140,166,168],{"class":167},"sgT6j","(",[140,170,171],{"class":150},"tf",[140,173,174],{"class":167},".",[140,176,178],{"class":177},"s3w6o","reduce_sum",[140,180,168],{"class":167},[140,182,171],{"class":150},[140,184,174],{"class":167},[140,186,187],{"class":150},"random",[140,189,174],{"class":167},[140,191,192],{"class":177},"normal",[140,194,195],{"class":167},"([",[140,197,199],{"class":198},"sYQis","1000",[140,201,202],{"class":167},",",[140,204,205],{"class":198}," 1000",[140,207,208],{"class":167},"])))\n",[45,210,211],{},[62,212],{"alt":64,"src":213},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fqevxkgqo.png",[215,216,218],"h3",{"id":217},"踩坑记录dll-load-failed找不到指定的模块",[88,219,220],{"code":220},"踩坑记录：(DLL load failed：找不到指定的模块)",[10,222,223],{},[45,224,225,226],{},"这里如果报如下图所示错误，是因为 vc++ 各种运行库缺失导致的，最简单的解决办法就是下载 vc++运行库合集，然后安装就好了，这里给出一个下载链接，不保证官方无毒，请自行斟酌使用。VC++ 运行库合集安装包下载地址：",[49,227,228],{"href":228,"rel":229},"http:\u002F\u002F8dx.pc6.com\u002Fwwb6\u002FWRYXKHJ2020.10.14.exe",[53],[45,231,232],{},[62,233],{"alt":64,"src":234},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Ftpjbcifj.png",[38,236,238],{"id":237},"_3-安装-keras-243","3. 安装 Keras 2.4.3",[56,240,241,254],{},[16,242,243,244,247,248,251],{},"继续在 terminal 中输入命令安装 Keras ",[88,245,246],{"code":246},"pip install keras -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\u002F"," ，阿里镜像源命令： ",[88,249,250],{"code":250},"pip install keras -i https:\u002F\u002Fmirrors.aliyun.com\u002Fpypi\u002Fsimple",[62,252],{"alt":64,"src":253},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fumijxxpv.png",[16,255,256],{},"输入一下 Python 代码，运行无报错则安装成功：",[132,258,260],{"className":134,"code":259,"language":136,"meta":64,"style":64},"import keras\n",[88,261,262],{"__ignoreMap":64},[140,263,264,266],{"class":142,"line":143},[140,265,147],{"class":146},[140,267,268],{"class":150}," keras\n",[45,270,271],{},[62,272],{"alt":64,"src":273},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fghvnsekw.png",[10,275,276],{},[45,277,278],{},"至此，TensorFlow 不带 GPU 的版本和 keras 已经安装完成，如需添加 GPU 支持，请继续如下操作",[38,280,282],{"id":281},"添加-gpu-支持","添加 GPU 支持",[10,284,285],{},[45,286,287],{},"以下内容全部基于普通笔记本电脑上的 NVIDIA GeForce MX150 独立显卡平台",[215,289,291],{"id":290},"_1-查找自己平台是否有-nvidia-独立显卡以及是否支持-cuda","1. 查找自己平台是否有 NVIDIA 独立显卡以及是否支持 CUDA",[56,293,294,304],{},[16,295,296,297,300,301],{},"打开任务管理器查看显卡型号，一般笔记本有核心显卡和独立显卡，找到有 NVIDIA 字样的一般就是英伟达的独立显卡了，接着复制显卡型号即图中的 ",[88,298,299],{"code":299},"NVIDIA GeForce MX150"," 到 NVIDIA 官网查询该型号是否支持 CUDA",[62,302],{"alt":64,"src":303},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fdwtlktph.png",[16,305,306,307],{},"可以看到如下图所示该显卡是支持 CUDA 的，所以可以给 TensorFlow 添加 GPU 支持",[62,308],{"alt":64,"src":309},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fdqmmsbex.png",[215,311,313],{"id":312},"_2-去官网下载-cuda101-以下的版本","2. 去官网下载 CUDA10.1 以下的版本",[10,315,316],{},[45,317,318],{},"目前 TensorFlow 最新版仅支持 10.1 即以下版本，不要安装最新的 CUDA11\ncuDNN 仅支持最高7.6版本",[56,320,321,331,334],{},[16,322,323,324,328],{},"CUDA 10.1 官网下载链接：",[49,325,326],{"href":326,"rel":327},"https:\u002F\u002Fdeveloper.nvidia.com\u002Fcuda-10.1-download-archive-base?target_os=Windows&target_arch=x86_64&target_version=10&target_type=exelocal",[53],[62,329],{"alt":64,"src":330},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Famybaynd.png",[16,332,333],{},"点击 Download 下载（点击一次可能会出现请求失败的 ERROR，重新点击就好了）",[16,335,336,337],{},"也可以点击这个链接直接下载，或者将该链接复制到迅雷里可以更快速的下载，实测迅雷能跑到 6mb\u002Fs：",[49,338,339],{"href":339,"rel":340},"https:\u002F\u002Fdeveloper.nvidia.com\u002Fcompute\u002Fcuda\u002F10.1\u002FProd\u002Flocal_installers\u002Fcuda_10.1.105_418.96_win10.exe",[53],[215,342,344],{"id":343},"_3-安装-cuda","3. 安装 CUDA",[56,346,347,353,359,365,371],{},[16,348,349,350],{},"双击打开安装包，点击 ok 加载缓存到临时目录",[62,351],{"alt":64,"src":352},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fsmcmhpxp.png",[16,354,355,356],{},"等待检查系统兼容性，然后点击同意并继续",[62,357],{"alt":64,"src":358},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fgdmumlba.png",[16,360,361,362],{},"选择自定义安装",[62,363],{"alt":64,"src":364},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fhqdjggft.png",[16,366,367,368],{},"只勾选如图所示的组件，其余一律不勾选",[62,369],{"alt":64,"src":370},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fjecarwdi.png",[16,372,373],{},"依次点击下一步，直至安装即可，默认安装位置尽量不要改动，后续要配置环境变量，如若更改，务必截图保存，以防忘记",[215,375,377],{"id":376},"_4-下载-cudnn-并配置环境变量","4. 下载 cuDNN 并配置环境变量",[56,379,380,390,393],{},[16,381,382,383,387],{},"点开此链接下载 cuDNN7.6 版本：",[49,384,385],{"href":385,"rel":386},"https:\u002F\u002Fdeveloper.nvidia.com\u002Frdp\u002Fcudnn-archive#a-collapse51b",[53],[62,388],{"alt":64,"src":389},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fjliivhsm.png",[16,391,392],{},"将下载的压缩包解压出来，把 cuda 文件夹复制到 C 盘根目录",[16,394,395],{},"将以下路径添加到系统环境变量中",[132,397,401],{"className":398,"code":399,"language":400,"meta":64,"style":64},"language-shell shiki shiki-themes catppuccin-latte one-dark-pro","C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1\\bin\nC:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1\\extras\\CUPTI\\lib64\nC:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1\\include\nC:\\cuda\\bin\n","shell",[88,402,403,447,488,519],{"__ignoreMap":64},[140,404,405,409,413,417,420,423,426,429,432,435,438,441,444],{"class":142,"line":143},[140,406,408],{"class":407},"seVD2","C:\\Program",[140,410,412],{"class":411},"sw_MA"," Files",[140,414,416],{"class":415},"sFWNk","\\N",[140,418,419],{"class":411},"VIDIA",[140,421,422],{"class":411}," GPU",[140,424,425],{"class":411}," Computing",[140,427,428],{"class":411}," Toolkit",[140,430,431],{"class":415},"\\C",[140,433,434],{"class":411},"UDA",[140,436,437],{"class":415},"\\v",[140,439,440],{"class":411},"10.1",[140,442,443],{"class":415},"\\b",[140,445,446],{"class":411},"in\n",[140,448,449,451,453,455,457,459,461,463,465,467,469,471,474,477,479,482,485],{"class":142,"line":160},[140,450,408],{"class":407},[140,452,412],{"class":411},[140,454,416],{"class":415},[140,456,419],{"class":411},[140,458,422],{"class":411},[140,460,425],{"class":411},[140,462,428],{"class":411},[140,464,431],{"class":415},[140,466,434],{"class":411},[140,468,437],{"class":415},[140,470,440],{"class":411},[140,472,473],{"class":415},"\\e",[140,475,476],{"class":411},"xtras",[140,478,431],{"class":415},[140,480,481],{"class":411},"UPTI",[140,483,484],{"class":415},"\\l",[140,486,487],{"class":411},"ib64\n",[140,489,491,493,495,497,499,501,503,505,507,509,511,513,516],{"class":142,"line":490},3,[140,492,408],{"class":407},[140,494,412],{"class":411},[140,496,416],{"class":415},[140,498,419],{"class":411},[140,500,422],{"class":411},[140,502,425],{"class":411},[140,504,428],{"class":411},[140,506,431],{"class":415},[140,508,434],{"class":411},[140,510,437],{"class":415},[140,512,440],{"class":411},[140,514,515],{"class":415},"\\i",[140,517,518],{"class":411},"nclude\n",[140,520,522],{"class":142,"line":521},4,[140,523,524],{"class":407},"C:\\cuda\\bin\n",[45,526,527,530,533,536],{},[62,528],{"alt":64,"src":529},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fmwdgxvhl.png",[62,531],{"alt":64,"src":532},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fvdgncjwe.png",[62,534],{"alt":64,"src":535},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fhtehyszh.png",[62,537],{"alt":64,"src":538},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Furpkuzqw.png",[215,540,542],{"id":541},"_5-运行验证是否支持-gpu","5. 运行验证是否支持 GPU",[56,544,545],{},[16,546,547],{},"在上文所建立的 pycharm 项目中，写入如下 Python 代码，执行",[132,549,550],{"className":134,"code":135,"language":136,"meta":64,"style":64},[88,551,552,562],{"__ignoreMap":64},[140,553,554,556,558,560],{"class":142,"line":143},[140,555,147],{"class":146},[140,557,151],{"class":150},[140,559,154],{"class":146},[140,561,157],{"class":150},[140,563,564,566,568,570,572,574,576,578,580,582,584,586,588,590,592,594],{"class":142,"line":160},[140,565,164],{"class":163},[140,567,168],{"class":167},[140,569,171],{"class":150},[140,571,174],{"class":167},[140,573,178],{"class":177},[140,575,168],{"class":167},[140,577,171],{"class":150},[140,579,174],{"class":167},[140,581,187],{"class":150},[140,583,174],{"class":167},[140,585,192],{"class":177},[140,587,195],{"class":167},[140,589,199],{"class":198},[140,591,202],{"class":167},[140,593,205],{"class":198},[140,595,208],{"class":167},[56,597,598],{"start":160},[16,599,600,601],{},"出现如下图所示提示即表示安装成功",[62,602],{"alt":64,"src":603},"https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fnholaknj.png",[38,605,607,608,610],{"id":606},"附录pip命令总结","附录：",[88,609,94],{"code":94},"命令总结",[612,613,614,630],"table",{},[615,616,617],"thead",{},[618,619,620,624],"tr",{},[621,622,623],"th",{},"pip 命令示例",[621,625,626],{},[627,628,629],"strong",{},"说明",[631,632,633,645,656,664,674,682,690,698,706],"tbody",{},[618,634,635,642],{},[636,637,638,639],"td",{},"pip download SomePackage",[140,640,641],{},"==version",[636,643,644],{},"下载扩展库的指定版本，不安装",[618,646,647,653],{},[636,648,649,650],{},"pip freeze ",[140,651,652],{},"> requirements.txt",[636,654,655],{},"以 requirements 的格式列出已安装模块",[618,657,658,661],{},[636,659,660],{},"pip list",[636,662,663],{},"列出当前已安装的所有模块",[618,665,666,671],{},[636,667,668,669],{},"pip install SomePackage",[140,670,641],{},[636,672,673],{},"在线安装 SomePackage 模块的指定版本",[618,675,676,679],{},[636,677,678],{},"pip install SomePackage.whl",[636,680,681],{},"通过 whl 文件离线安装扩展库",[618,683,684,687],{},[636,685,686],{},"pip install package1 package2 …",[636,688,689],{},"依次（在线）安装 package1、package2 等扩展模块",[618,691,692,695],{},[636,693,694],{},"pip install -r requirements.txt",[636,696,697],{},"安装 requirements.txt 文件中指定的扩展库",[618,699,700,703],{},[636,701,702],{},"pip install –upgrade SomePackage",[636,704,705],{},"升级 SomePackage 模块",[618,707,708,713],{},[636,709,710,711],{},"pip uninstall SomePackage",[140,712,641],{},[636,714,715],{},"卸载 SomePackage 模块的指定版本",[717,718,719],"style",{},"html pre.shiki code .sSWcl, html code.shiki .sSWcl{--shiki-default:#8839EF;--shiki-dark:#C678DD}html pre.shiki code .sa2x1, html code.shiki .sa2x1{--shiki-default:#4C4F69;--shiki-dark:#ABB2BF}html pre.shiki code .sk-YW, html code.shiki .sk-YW{--shiki-default:#FE640B;--shiki-default-font-style:italic;--shiki-dark:#56B6C2;--shiki-dark-font-style:inherit}html pre.shiki code .sgT6j, html code.shiki .sgT6j{--shiki-default:#7C7F93;--shiki-dark:#ABB2BF}html pre.shiki code .s3w6o, html code.shiki .s3w6o{--shiki-default:#1E66F5;--shiki-dark:#61AFEF}html pre.shiki code .sYQis, html code.shiki .sYQis{--shiki-default:#FE640B;--shiki-dark:#D19A66}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html pre.shiki code .seVD2, html code.shiki .seVD2{--shiki-default:#1E66F5;--shiki-default-font-style:italic;--shiki-dark:#61AFEF;--shiki-dark-font-style:inherit}html pre.shiki code .sw_MA, html code.shiki .sw_MA{--shiki-default:#40A02B;--shiki-dark:#98C379}html pre.shiki code .sFWNk, html code.shiki .sFWNk{--shiki-default:#EA76CB;--shiki-dark:#56B6C2}",{"title":64,"searchDepth":521,"depth":521,"links":721},[722,723,726,727,734],{"id":40,"depth":160,"text":41},{"id":74,"depth":160,"text":75,"children":724},[725],{"id":217,"depth":490,"text":220},{"id":237,"depth":160,"text":238},{"id":281,"depth":160,"text":282,"children":728},[729,730,731,732,733],{"id":290,"depth":490,"text":291},{"id":312,"depth":490,"text":313},{"id":343,"depth":490,"text":344},{"id":376,"depth":490,"text":377},{"id":541,"depth":490,"text":542},{"id":606,"depth":160,"text":735},"附录：pip命令总结",[737],"技术","2020-10-06 11:18:24","本文详细介绍了在 Windows 10 系统下使用 PyCharm 搭建深度学习环境的完整流程。首先基于 Anaconda 创建 Python 项目环境，随后通过 PyCharm 内置终端使用清华或阿里镜像源安装 TensorFlow 2.3.1 和 Keras 2.4.3，并验证安装成功。针对常见 DLL 加载失败问题，建议安装 VC++ 运行库合集解决。若需启用 GPU 加速，文章指导用户确认 NVIDIA 显卡（如 MX150）对 CUDA 的支持，下载并安装兼容的 CUDA 10.1 和 cuDNN 7.6，配置相应系统环境变量，最终通过 TensorFlow 代码验证 GPU 是否正常调用。文末附常用 pip 命令速查表，便于环境管理。",false,"md","https:\u002F\u002Ffile.dhbxs.top\u002F2025\u002F10\u002Fsukseffe.webp",{"slots":744},{},true,"\u002Fposts\u002F37561f92",null,{"text":749,"minutes":750,"time":751,"words":752},"6 min read",5.03,301800,1006,0,{"title":5,"description":739},{"loc":746},"posts\u002F2020\u002F利用PyCharm简化搭建深度学习环境(新手排坑)",[758,759],"深度学习","PyCharm","tech","8ViaOk5oellLPfb5OA9b_ErRe1GXC4PzTlPPvSGz7dY",[763,769],{"title":764,"path":765,"stem":766,"date":767,"type":768,"children":-1},"Win10简洁之道","\u002Fposts\u002Ffc1be4e5","posts\u002F2020\u002FWin10简洁之道","2020-03-31 11:12:04","story",{"title":770,"path":771,"stem":772,"date":773,"type":760,"children":-1},"Hadoop集群搭建（1主2从）","\u002Fposts\u002Fd06540f2","posts\u002F2024\u002FHadoop集群搭建（1主2从）","2024-04-18 06:26:21",1776079556375]