{"id":60499,"date":"2021-07-05T21:27:32","date_gmt":"2021-07-05T12:27:32","guid":{"rendered":"https:\/\/smilegate.ai\/?p=60499"},"modified":"2021-07-05T21:48:15","modified_gmt":"2021-07-05T12:48:15","slug":"tensorflow-vs-pytorch","status":"publish","type":"post","link":"https:\/\/smilegate.ai\/cn\/2021\/07\/05\/tensorflow-vs-pytorch\/","title":{"rendered":"Tensorflow VS Pytorch"},"content":{"rendered":"

[\u670d\u52a1\u5f00\u53d1\u56e2\u961f\uff0cHyun-Jun Han]<\/p>\n\n\n\n

\u4ecb\u7ecd<\/strong><\/p>\n\n\n\n

Tensorflow \u548c PyTorch \u662f\u6700\u6d41\u884c\u7684\u673a\u5668\u5b66\u4e60\u5e93\u3002\u5f00\u53d1 AI \u670d\u52a1\u7684\u5f00\u53d1\u4eba\u5458\u6216\u7814\u7a76\u6a21\u578b\u7684\u7814\u7a76\u4eba\u5458\u5728\u5f00\u59cb\u65f6\u5e38\u5e38\u9762\u4e34\u9009\u62e9\u56fe\u4e66\u9986\u7684\u56f0\u96be\u3002\u6b64\u5916\uff0c\u968f\u7740\u5e93\u7248\u672c\u7684\u5347\u7ea7\uff0c\u6709\u5f88\u591a\u53d8\u5316\uff0c\u6240\u4ee5\u6211\u4eec\u60f3\u63d0\u4f9b\u4e00\u4e2a\u9009\u62e9\u5e93\u7684\u6307\u5357\u3002<\/p>\n\n\n\n

\u8c37\u6b4c \u2013 Tensorflow<\/strong><\/p>\n\n\n\n

Tensorflow \u7531 Google Brain \u56e2\u961f\u521b\u5efa\uff0c\u7528\u4e8e\u7814\u7a76\u548c\u4ea7\u54c1\u5f00\u53d1\uff0c\u517c\u5bb9\u8bed\u8a00\u5305\u62ec C++\u3001Python\u3001JavaScript \u548c Swift\u3002\u5728 1.x \u548c 2.x \u7248\u672c\u4e4b\u540e\uff0c\u5df2\u7ecf\u53d1\u5e03\u4e86\u7a33\u5b9a\u7248\u672c 2.4\u3002 1.x \u548c 2.x \u7248\u672c\u7684\u533a\u522b\u5728\u4e8e\u51e0\u4e4e\u4e0d\u9700\u8981\u4f7f\u7528 Session\uff0c\u58f0\u660e\u6700\u5c0f\u5316\u548c Keras \u88ab\u5185\u5316\u4e86\u3002\u6b64\u5916\uff0c\u8fd8\u901a\u8fc7\u529f\u80fd\u5316\u63d0\u9ad8\u4e86\u901f\u5ea6\u3002\u8fd8\u6709\u8bb8\u591a\u5176\u4ed6\u66f4\u6539\uff0c\u5305\u62ec Tensorflow-Lite - \u79fb\u52a8\/\u5d4c\u5165\u652f\u6301\u3001TensorflowJS - \u6d4f\u89c8\u5668\u3001Node.js \u652f\u6301\u7b49\u3002\u4f46\u662f\uff0c\u8fd9\u79cd\u66f4\u6539\u5bfc\u81f4\u5fc5\u987b\u66f4\u6539\u5df2\u7ecf\u4f7f\u7528\u5b83\u7684\u7528\u6237\u7684\u4ee3\u7801\u7684\u4e0d\u4fbf\u3002\u5f53\u7136\uff0c\u5b83\u652f\u6301\u8f6c\u6362\u6a21\u5757\uff0c\u4f46\u4e5f\u6709bug\uff0c\u6240\u4ee5\u5f88\u591a\u7528\u6237\u79bb\u5f00\u4e86\u3002<\/p>\n\n\n\n

\u8138\u4e66 \u2013 PyTorch<\/strong><\/p>\n\n\n\n

PyTorch \u662f Facebook AI Research Lab \u4e3a\u7814\u7a76\u76ee\u7684\u5f00\u53d1\u7684\u5e93\u3002\u517c\u5bb9\u8bed\u8a00\u5305\u62ecC++\u548cPython\u3002\u4e00\u5f00\u59cb\uff0c\u793e\u533a\u89c4\u6a21\u8f83\u5c0f\uff0c\u4f46\u53d1\u5c55\u8fc5\u901f\uff0c\u5728\u4f1a\u8bae\u8bba\u6587\u4e2d\u8d85\u8fc7\u4e86 Tensorflow\u3002\u8bb8\u591a\u7814\u7a76\u9886\u57df\u7684\u7528\u6237\u8f6c\u5411 PyTorch \u7684\u539f\u56e0\u662f\u5b83\u53ef\u4ee5\u4f7f\u7528\u52a8\u6001\u56fe\u3002\u8fd9\u662f\u56e0\u4e3a\u5b83\u6709\u5229\u4e8e RNN\u3001CNN \u548c GAN \u7814\u7a76\uff0c\u56e0\u4e3a\u53ef\u4ee5\u901a\u8fc7\u52a8\u6001\u56fe\u5b9e\u65f6\u66ff\u6362\u6570\u636e\u6765\u6bd4\u8f83\u6570\u636e\u3002\u6b64\u5916\uff0c\u6211\u8ba4\u4e3a\u4ee3\u7801\u5df2\u7ecf\u663e\u7740\u589e\u52a0\uff0c\u56e0\u4e3a\u4e0e Tensorflow \u76f8\u6bd4\u96be\u5ea6\u76f8\u5bf9\u5bb9\u6613\u3002<\/p>\n\n\n\n

Tensorflow VS PyTorch<\/strong><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u8fd9\u662f\u4eba\u5de5\u667a\u80fd\u9886\u57df\u5404\u4e2a\u4f1a\u8bae\u7684\u8bba\u6587\u56fe\u8868\u3002\u53ef\u4ee5\u770b\u51fa\uff0cPyTorch \u7684\u8bba\u6587\u7528\u5b9e\u7ebf\u8868\u793a\u7684\u8d8b\u52bf\u9ad8\u4e8e\u7528\u865a\u7ebf\u8868\u793a\u7684 Tensorflow\u3002\u6211\u8ba4\u4e3a PyTorch \u5728\u7814\u7a76\u4f1a\u8bae\u8bba\u6587\u65b9\u9762\u66f4\u5177\u4f18\u52bf\u3002<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/td>\u5f20\u91cf\u6d41<\/td>\u706b\u70ac<\/td><\/tr>
Github \u5b58\u50a8\u5e93<\/td>108k<\/td>68k<\/td><\/tr>
\u53ef\u89c6\u5316<\/td>\u5f20\u91cf\u677f<\/td>\u7ef4\u591a\u59c6
\u5f20\u91cf\u677f [1.2 Pytorch]<\/td><\/tr>
\u56fe\u5b9a\u4e49<\/td>\u9759\u6001\u56fe\uff0c
\u52a8\u6001\u56fe [tensorflow 2.0]<\/td>		\u52a8\u6001\u56fe<\/td><\/tr>
\u8c03\u8bd5<\/td>tfdbg<\/td>Python \u6807\u51c6\u8c03\u8bd5\u5668<\/td><\/tr>
\u6a21\u578b\u90e8\u7f72<\/td>Tensorflow \u670d\u52a1<\/td>\u706b\u70ac\u670d\u52a1<\/td><\/tr>
\u79fb\u52a8\u652f\u6301<\/td>\u54e6<\/td>\u54e6<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

\u4ece\u529f\u80fd\u4e0a\u6bd4\u8f83\u6bcf\u4e2a\u5e93\u65f6\uff0c\u5b83\u4eec\u4f3c\u4e4e\u968f\u7740\u7248\u672c\u7684\u589e\u52a0\u800c\u8d8a\u6765\u8d8a\u76f8\u540c\u3002\u8d77\u521d\uff0c\u4e24\u4e2a\u5e93\u4e4b\u95f4\u5b58\u5728\u8bb8\u591a\u5dee\u5f02\uff0c\u4f46\u968f\u7740\u65f6\u95f4\u7684\u63a8\u79fb\uff0c\u5dee\u5f02\u663e\u7740\u7f29\u5c0f\uff0c\u56e0\u4e3a\u5b83\u4eec\u5177\u6709\u66f4\u591a\u5f7c\u6b64\u7684\u4f18\u70b9\u3002 <\/p>\n\n\n\n

\u5c3d\u7ba1\u5982\u6b64\uff0c\u8fd9\u4e24\u4e2a\u5e93\u7684\u5c40\u9650\u6027\u4ecd\u7136\u5b58\u5728\u3002 Tensorflow \u6784\u5efa\u52a8\u6001\u56fe\u8fd8\u662f\u6bd4\u8f83\u56f0\u96be\u7684\uff0c\u8c03\u8bd5\u4e5f\u9700\u8981\u8bad\u7ec3\u3002 PyTorch\u76ee\u524d\u5f00\u6e90\u5f88\u5c11\uff0c\u96be\u4ee5\u652f\u6301\u5176\u4ed6\u73af\u5883\uff0c\u96be\u4ee5\u90e8\u7f72\uff0c\u6700\u91cd\u8981\u7684\u662f\u96be\u4ee5\u6ee1\u8db3\u884c\u4e1a\u6240\u9700\u7684\u5404\u79cd\u7ea6\u675f\u3002\u8003\u8651\u5230\u8fd9\u4e00\u70b9\uff0c\u5982\u679c\u60a8\u5728\u6e05\u695a\u4e86\u89e3\u91cd\u70b9\u662f\u7528\u4e8e\u7814\u7a76\u76ee\u7684\u8fd8\u662f\u63d0\u4f9b\u4ea7\u54c1\u548c\u670d\u52a1\u4e4b\u540e\u7ee7\u7eed\u8fdb\u884c\u9879\u76ee\uff0c\u5c06\u6709\u52a9\u4e8e\u9009\u62e9\u56fe\u4e66\u9986\u3002<\/p>\n\n\n\n

\u7ed3\u8bba<\/strong><\/p>\n\n\n\n

Tensorflow - \u7528\u4e8e\u4ea7\u54c1\u6784\u5efa\u76ee\u7684\u3002\u6d4f\u89c8\u5668\u3001Node.js\u3001\u5d4c\u5165\u5f0f\u3001\u79fb\u52a8\u73af\u5883\u652f\u6301\u3002\u6613\u4e8e\u90e8\u7f72\u3002\u81ea\u7136\u8bed\u8a00\u5904\u7406\u7b49<\/p>\n\n\n\n

Pytorch - \u7528\u4e8e\u7814\u7a76\u76ee\u7684\u3002\u719f\u6089 Python \u7684\u73af\u5883\u3002\u96be\u5ea6\u76f8\u5bf9\u8f83\u4f4e\u3002 RNN\u3001CNN\u3001GAN \u7b49\u3002<\/p>\n\n\n\n

Tensorflow \u548c Pytorch \u90fd\u662f\u5f3a\u5927\u7684\u673a\u5668\u5b66\u4e60\u5e93\u3002\u5982\u679c\u60a8\u6839\u636e\u81ea\u5df1\u7684\u76ee\u7684\u9009\u62e9\u56fe\u4e66\u9986\uff0c\u6211\u8ba4\u4e3a\u4f1a\u8fdb\u884c\u66f4\u6709\u6210\u6548\u7684\u7814\u7a76\/\u5f00\u53d1\u3002<\/p>\n\n\n\n

\u53c2\u8003<\/p>\n\n\n\n

https:\/\/towardsdatascience.com\/pytorch-vs-tensorflow-2021-d403504d7bc3<\/a><\/p>\n\n\n\n

https:\/\/builtin.com\/data-science\/pytorch-vs-tensorflow<\/a><\/p>\n\n\n\n

https:\/\/www.imaginarycloud.com\/blog\/pytorch-vs-tensorflow\/<\/a><\/p>\n

<\/span><\/div>","protected":false},"excerpt":{"rendered":"

[\uc11c\ube44\uc2a4\uac1c\ubc1c\ud300 \ud55c\ud604\uc900] \uc11c\ub860 Tensorflow\uc640 PyTorch\ub294 \uae30\uacc4 \ud559\uc2b5\uc5d0 \uc788\uc5b4\uc11c \uac00\uc7a5 \uc778\uae30 \uc788\ub294 \ub77c\uc774\ube0c\ub7ec\ub9ac\ub4e4\uc785\ub2c8\ub2e4. AI \uc11c\ube44\uc2a4\ub97c \uac1c\ubc1c\ud558\ub294 \uac1c\ubc1c\uc790\ub098 \ubaa8\ub378\uc744 \uc5f0\uad6c\ud558\ub294 \uc5f0\uad6c\uc790\ub4e4\uc740 \ucd08\uae30\uc5d0 \ub77c\uc774\ube0c\ub7ec\ub9ac \uc120\ud0dd\uc5d0 \uc788\uc5b4\uc11c \uc5b4\ub824\uc6c0\uc744 \uacaa\uace4 \ud569\ub2c8\ub2e4. \ub610\ud55c \ub77c\uc774\ube0c\ub7ec\ub9ac \ubc84\uc804\uc774 \uc62c\ub77c\uac10\uc5d0 \ub530\ub77c \ubcc0\uacbd\ub41c \uc0ac\ud56d\uc774 \ub9ce\uc544\uc84c\uae30 \ub54c\ubb38\uc5d0 \ub77c\uc774\ube0c\ub7ec\ub9ac \uc120\ud0dd\ud558\ub294\ub370 \uc788\uc5b4 \uac00\uc774\ub4dc\ub97c \uc81c\uacf5\ud558\uace0\uc790 \ud569\ub2c8\ub2e4. Google – Tensorflow Tensorflow\ub294 Google Brain Team\uc774 \uc5f0\uad6c\uc640 \uc81c\ud488 \uac1c\ubc1c\uc744 \ubaa9\uc801\uc73c\ub85c \ub9cc\ub4e4\uc5c8\uc73c\uba70 \ud638\ud658 \uc5b8\uc5b4\ub85c\ub294…<\/p>\n

<\/span><\/div>","protected":false},"author":1,"featured_media":60510,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"rank_math_lock_modified_date":false,"footnotes":""},"categories":[19],"tags":[464,500],"class_list":["post-60499","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tech04","tag-tensorflow","tag-pytorch","category-19","description-off"],"_links":{"self":[{"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/posts\/60499","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/comments?post=60499"}],"version-history":[{"count":8,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/posts\/60499\/revisions"}],"predecessor-version":[{"id":60515,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/posts\/60499\/revisions\/60515"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/media\/60510"}],"wp:attachment":[{"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/media?parent=60499"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/categories?post=60499"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/smilegate.ai\/cn\/wp-json\/wp\/v2\/tags?post=60499"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}