{"id":5067,"date":"2024-11-12T14:10:44","date_gmt":"2024-11-12T06:10:44","guid":{"rendered":"https:\/\/nullthought.net\/?p=5067"},"modified":"2024-12-25T14:30:34","modified_gmt":"2024-12-25T06:30:34","slug":"%e6%80%9d%e7%bb%b4%e9%93%becot%e8%a6%81%e6%9c%89%e9%80%89%e6%8b%a9%e5%9c%b0%e4%bd%bf%e7%94%a8","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=5067","title":{"rendered":"\u601d\u7ef4\u94feCoT\u8981\u6709\u9009\u62e9\u5730\u4f7f\u7528"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">\u8bba\u6587<strong><a href=\"https:\/\/arxiv.org\/abs\/2410.21333\" target=\"_blank\" rel=\"noreferrer noopener\">Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse<\/a><\/strong>\u63a2\u8ba8\u4e86<strong>\u601d\u7ef4\u94fe<\/strong>\uff08\u6216\u79f0\u201c\u9010\u6b65\u63a8\u7406\u201d\uff0cChain-of-Thought, CoT\uff09\u5728\u67d0\u4e9b\u4efb\u52a1\u4e2d\u53ef\u80fd\u4f1a\u5bfc\u81f4\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u6027\u80fd\u4e0b\u964d\u7684\u60c5\u51b5\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u968f\u7740\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u548c\u591a\u6a21\u6001\u6a21\u578b\uff08LMM\uff09\u7684\u53d1\u5c55\uff0c\u601d\u7ef4\u94feCoT\u9010\u6e10\u6210\u4e3a\u8fd9\u4e9b\u6a21\u578b\u4e2d\u4e00\u79cd\u666e\u904d\u4f7f\u7528\u7684\u63a8\u7406\u65b9\u5f0f\u3002\u8fd9\u79cd\u65b9\u6cd5\u901a\u8fc7\u5728\u56de\u7b54\u4e4b\u524d\u8fdb\u884c\u9010\u6b65\u7684\u903b\u8f91\u63a8\u5bfc\uff0c<a href=\"https:\/\/nullthought.net\/?p=4744\" target=\"_blank\" rel=\"noreferrer noopener\">\u4f7f\u5f97\u6a21\u578b\u53ef\u4ee5\u66f4\u597d\u5730\u5904\u7406\u9700\u8981\u590d\u6742\u63a8\u7406\u7684\u4efb\u52a1\uff0c\u4f8b\u5982\u6570\u5b66\u8fd0\u7b97\u548c\u7b26\u53f7\u903b\u8f91<\/a>\u3002\u7136\u800c\uff0c\u8fd9\u79cd\u65b9\u6cd5\u5e76\u4e0d\u603b\u662f\u80fd\u63d0\u5347\u6a21\u578b\u7684\u6027\u80fd\u3002\u5728\u67d0\u4e9b\u60c5\u51b5\u4e0b\uff0c\u601d\u7ef4\u94feCoT\u751a\u81f3\u4f1a\u5bfc\u81f4\u6a21\u578b\u6027\u80fd\u4e0b\u964d\u3002\u56e0\u6b64\uff0c\u672c\u8bba\u6587\u7684\u7814\u7a76\u91cd\u70b9\u5728\u4e8e\u8bc6\u522b\u548c\u5206\u6790\u54ea\u4e9b\u4efb\u52a1\u4e2d\u601d\u7ef4\u94feCoT\u5bf9\u6a21\u578b\u4ea7\u751f\u4e86\u8d1f\u9762\u5f71\u54cd\uff0c\u5c24\u5176\u662f\u501f\u9274\u8ba4\u77e5\u5fc3\u7406\u5b66\u4e2d\u5173\u4e8e\u4eba\u7c7b\u5728\u67d0\u4e9b\u4efb\u52a1\u4e2d\u7ecf\u8fc7\u6df1\u601d\u719f\u8651\u540e\u8868\u73b0\u53cd\u800c\u53d8\u5dee\u7684\u73b0\u8c61\u3002\u901a\u8fc7\u5bf9\u6bd4\u4eba\u7c7b\u548c\u6a21\u578b\u7684\u8ba4\u77e5\u8fc7\u7a0b\uff0c\u7814\u7a76\u5e0c\u671b\u627e\u5230CoT\u5f71\u54cd\u6a21\u578b\u8868\u73b0\u7684\u89c4\u5f8b\u548c\u7279\u5f81\uff0c\u4ece\u800c\u4e3a\u672a\u6765\u6a21\u578b\u63a8\u7406\u7b56\u7565\u7684\u9009\u62e9\u63d0\u4f9b\u4f9d\u636e\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8bba\u6587\u4f5c\u8005\u4e3aRyan Liu, Jiayi Geng, Addison J. Wu, Ilia Sucholutsky, Tania Lombrozo, Thomas L. Griffiths\uff0c\u6765\u81eaPrinceton University\u548cNew York University\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"761\" height=\"630\" src=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-10.png\" alt=\"\" class=\"wp-image-5068\" srcset=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-10.png 761w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-10-300x248.png 300w\" sizes=\"auto, (max-width: 761px) 100vw, 761px\" \/><figcaption class=\"wp-element-caption\"><strong><a href=\"https:\/\/arxiv.org\/abs\/2410.21333\" target=\"_blank\" rel=\"noreferrer noopener\">Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse<\/a><\/strong><\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u8bba\u6587\u63d0\u51fa\u4e86\u4e24\u4e2a\u5173\u952e\u7684\u7814\u7a76\u5047\u8bbe\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u5f53\u4efb\u52a1\u4e2d\u201c\u53e3\u5934\u601d\u8003\u201d\u6216\u6df1\u601d\u719f\u8651\u4f1a\u5f71\u54cd\u4eba\u7c7b\u8868\u73b0\u65f6\uff0c\u7c7b\u4f3c\u7684\u63a8\u7406\u65b9\u5f0f\uff08\u5982\u601d\u7ef4\u94feCoT\uff09\u53ef\u80fd\u4e5f\u4f1a\u5bf9\u6a21\u578b\u7684\u8868\u73b0\u4ea7\u751f\u8d1f\u9762\u5f71\u54cd\u3002<\/li>\n\n\n\n<li>\u8fd9\u4e9b\u5f71\u54cd\u662f\u57fa\u4e8e\u4efb\u52a1\u7684\u6027\u8d28\u53ca\u4eba\u7c7b\u548c\u6a21\u578b\u5728\u4efb\u52a1\u4e2d\u7684\u9650\u5236\u7684\u76f8\u4f3c\u6027\uff0c\u56e0\u6b64\uff0c\u6a21\u578b\u5728\u67d0\u4e9b\u60c5\u51b5\u4e0b\u53ef\u80fd\u4f1a\u590d\u73b0\u4eba\u7c7b\u7684\u5931\u8d25\u6a21\u5f0f\uff0c\u800c\u5728\u5176\u4ed6\u60c5\u51b5\u4e0b\u4e0d\u4f1a\u3002<\/li>\n<\/ol>\n\n\n\n<p class=\"wp-block-paragraph\">\u4e3a\u4e86\u9a8c\u8bc1\u8fd9\u4e9b\u5047\u8bbe\uff0c\u4f5c\u8005\u8bbe\u8ba1\u4e86\u516d\u7c7b\u5b9e\u9a8c\uff0c\u5176\u4e2d\u4e09\u7c7b\u4efb\u52a1\u6ee1\u8db3\u4eba\u7c7b\u8868\u73b0\u53d7\u635f\u7684\u6761\u4ef6\uff0c\u540c\u65f6\u9002\u7528\u4e8e\u8bed\u8a00\u6a21\u578b\uff1b\u53e6\u5916\u4e09\u7c7b\u4efb\u52a1\u5c3d\u7ba1\u4eba\u7c7b\u8868\u73b0\u53d8\u5dee\uff0c\u4f46\u7531\u4e8e\u6a21\u578b\u4e0e\u4eba\u7c7b\u7684\u8ba4\u77e5\u65b9\u5f0f\u4e0d\u540c\uff0c\u601d\u7ef4\u94feCoT\u5bf9\u6a21\u578b\u7684\u5f71\u54cd\u53ef\u80fd\u4e0d\u76f8\u540c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5177\u4f53\u6765\u8bf4\uff0c\u4f5c\u8005\u9009\u62e9\u4e86\u4ee5\u4e0b\u516d\u7c7b\u4efb\u52a1\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u9690\u5f0f\u7edf\u8ba1\u5b66\u4e60\uff08Implicit Statistical Learning\uff09<\/li>\n\n\n\n<li>\u9762\u90e8\u8bc6\u522b\uff08Facial Recognition\uff09<\/li>\n\n\n\n<li>\u5904\u7406\u5305\u542b\u4f8b\u5916\u7684\u5206\u7c7b\uff08Classification with Exceptions\uff09<\/li>\n\n\n\n<li>\u903b\u8f91\u4e0d\u4e00\u81f4\u6027\u68c0\u6d4b\uff08Logical Inconsistency Detection\uff09<\/li>\n\n\n\n<li>\u7a7a\u95f4\u76f4\u89c9\u4efb\u52a1\uff08Spatial Intuition Task\uff09<\/li>\n\n\n\n<li>\u805a\u5408\u591a\u7ef4\u7279\u5f81\u7684\u51b3\u7b56\u4efb\u52a1\uff08Aggregating Features for Decision Making\uff09<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u4e00\uff1a\u9690\u5f0f\u7edf\u8ba1\u5b66\u4e60<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u9690\u5f0f\u7edf\u8ba1\u5b66\u4e60\u662f\u6307\u4eba\u7c7b\u5728\u6ca1\u6709\u663e\u5f0f\u5b66\u4e60\u89c4\u5219\u7684\u60c5\u51b5\u4e0b\uff0c\u80fd\u591f\u4ece\u4e00\u7cfb\u5217\u8f93\u5165\u4e2d<strong>\u65e0\u610f\u8bc6<\/strong>\u5730\u5b66\u4e60\u51fa\u5176\u5185\u5728\u7684\u7edf\u8ba1\u89c4\u5f8b\u3002\u4f8b\u5982\uff0c\u5728\u4eba\u5de5\u6587\u6cd5\u5b66\u4e60\u4efb\u52a1\u4e2d\uff0c\u4eba\u4eec\u88ab\u8981\u6c42\u5bf9\u5b57\u6bcd\u5e8f\u5217\u8fdb\u884c\u5206\u7c7b\uff0c\u5373\u5224\u65ad\u5b83\u4eec\u662f\u5426\u7b26\u5408\u67d0\u79cd\u89c4\u5219\u3002\u7814\u7a76\u8868\u660e\uff0c\u4eba\u7c7b\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\u5f80\u5f80\u65e0\u6cd5\u7528\u8bed\u8a00\u63cf\u8ff0\u5b66\u4e60\u5230\u7684\u89c4\u5219\uff0c\u4f46\u5374\u80fd\u901a\u8fc7<strong>\u76f4\u89c9<\/strong>\u5224\u65ad\u51fa\u54ea\u4e9b\u5e8f\u5217\u662f\u5408\u6cd5\u7684\u3002\u7136\u800c\uff0c\u5982\u679c\u88ab\u8981\u6c42\u7528\u8bed\u8a00\u63cf\u8ff0\u8fd9\u4e9b\u89c4\u5219\uff0c\u4eba\u7c7b\u7684\u5206\u7c7b\u8868\u73b0\u4f1a\u663e\u8457\u4e0b\u964d\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">2. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5728\u672c\u7814\u7a76\u4e2d\uff0c\u4f5c\u8005\u6a21\u62df\u4e86\u8fd9\u79cd\u9690\u5f0f\u5b66\u4e60\u7684\u60c5\u5883\uff0c\u751f\u6210\u4e86100\u4e2a\u7531\u6709\u9650\u72b6\u6001\u81ea\u52a8\u673a\uff08FSG\uff09\u751f\u6210\u7684\u5b57\u6bcd\u5e8f\u5217\uff0c\u5e76\u8981\u6c42\u591a\u4e2a\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08\u5982OpenAI o1-preview\u3001GPT-4o\u3001Claude 3 Opus\u7b49\uff09\u5728\u96f6\u6837\u672c\uff08zero-shot\uff09\u548c\u601d\u7ef4\u94fe\uff08CoT\uff09\u6761\u4ef6\u4e0b\u8fdb\u884c\u5206\u7c7b\u3002\u6bcf\u4e2a\u5206\u7c7b\u95ee\u9898\u5305\u542b15\u4e2a\u8bad\u7ec3\u6837\u672c\u548c44\u4e2a\u6d4b\u8bd5\u6837\u672c\uff0c\u5176\u4e2d\u4e00\u534a\u662f\u5408\u6cd5\u5e8f\u5217\uff0c\u53e6\u4e00\u534a\u5219\u901a\u8fc7\u5bf9\u5408\u6cd5\u5e8f\u5217\u8fdb\u884c\u968f\u673a\u4fee\u6539\u5f97\u5230\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u7ed3\u679c\u663e\u793a\uff0c\u601d\u7ef4\u94feCoT\u663e\u8457\u964d\u4f4e\u4e86\u6a21\u578b\u7684\u8868\u73b0\u3002\u4f8b\u5982\uff0cOpenAI o1-preview\u6a21\u578b\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u7684\u5206\u7c7b\u51c6\u786e\u7387\u6bd4GPT-4o\u96f6\u6837\u672c\u6761\u4ef6\u4e0b\u964d\u4e8636.3%\u3002\u8fd9\u4e00\u7ed3\u679c\u8868\u660e\uff0c\u601d\u7ef4\u94feCoT\u5bfc\u81f4\u6a21\u578b\u503e\u5411\u4e8e\u901a\u8fc7\u663e\u5f0f\u7684\u8bed\u8a00\u903b\u8f91\u6765\u63a8\u5bfc\u9690\u5f0f\u6a21\u5f0f\uff0c\u800c\u8fd9\u79cd\u65b9\u6cd5\u5e76\u4e0d\u9002\u7528\u4e8e\u6355\u6349\u590d\u6742\u7684\u7edf\u8ba1\u89c4\u5f8b\uff0c\u53cd\u800c\u5f71\u54cd\u4e86\u6a21\u578b\u5bf9\u6570\u636e\u7684\u5f52\u7eb3\u80fd\u529b\u3002\u8fd9\u4e0e\u4eba\u7c7b\u5728\u8fd9\u79cd\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\u975e\u5e38\u7c7b\u4f3c\uff0c\u5373\u8bed\u8a00\u63cf\u8ff0\u7684\u5e72\u9884\u53ef\u80fd\u4f1a\u7834\u574f\u5bf9\u9690\u5f0f\u89c4\u5f8b\u7684\u611f\u77e5\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u4e8c\uff1a\u9762\u90e8\u8bc6\u522b<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u9762\u90e8\u8bc6\u522b\u4efb\u52a1\u901a\u5e38\u6d89\u53ca\u89c6\u89c9\u7279\u5f81\u7684\u8bc6\u522b\u548c\u5339\u914d\u3002\u5728\u4eba\u7c7b\u7684\u8ba4\u77e5\u5b9e\u9a8c\u4e2d\uff0c\u8981\u6c42\u53c2\u4e0e\u8005\u5bf9\u770b\u5230\u7684\u9762\u90e8\u8fdb\u884c\u53e3\u5934\u63cf\u8ff0\uff0c\u5f80\u5f80\u4f1a\u5bfc\u81f4\u5176\u4e4b\u540e\u7684\u9762\u90e8\u8bc6\u522b\u8868\u73b0\u4e0b\u964d\uff0c\u8fd9\u88ab\u79f0\u4e3a\u201c\u8bed\u8a00\u906e\u853d\u6548\u5e94\u201d\uff08Verbal Overshadowing\uff09\u3002\u8fd9\u662f\u56e0\u4e3a\u8bed\u8a00\u5728\u63cf\u8ff0\u89c6\u89c9\u7ec6\u8282\u65b9\u9762\u5b58\u5728\u5c40\u9650\u6027\uff0c\u96be\u4ee5\u6709\u6548\u6355\u6349\u9762\u90e8\u7684\u72ec\u7279\u7279\u5f81\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">2. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u4f5c\u8005\u4f7f\u7528\u4e86\u7531\u751f\u6210\u6a21\u578b\u751f\u6210\u7684\u4eba\u8138\u56fe\u50cf\u6570\u636e\u96c6\uff0c\u5305\u542b\u4e86\u4e0d\u540c\u6027\u522b\u3001\u79cd\u65cf\u3001\u53d1\u8272\u3001\u773c\u775b\u989c\u8272\u7b49\u7279\u5f81\u7684\u9762\u5b54\u3002\u6bcf\u4e2a\u95ee\u9898\u7531\u4e00\u5f20\u76ee\u6807\u4eba\u8138\u548c\u4e94\u5f20\u5019\u9009\u4eba\u8138\u7ec4\u6210\uff0c\u5176\u4e2d\u53ea\u6709\u4e00\u5f20\u4e0e\u76ee\u6807\u4eba\u8138\u5339\u914d\u3002\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\uff08\u5982GPT-4o\u3001Claude 3 Opus\u7b49\uff09\u88ab\u8981\u6c42\u5728\u96f6\u6837\u672c\u548c\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u8fdb\u884c\u9762\u90e8\u5339\u914d\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u6240\u6709\u6d4b\u8bd5\u7684\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u7684\u8868\u73b0\u5747\u6709\u6240\u4e0b\u964d\u3002\u4f8b\u5982\uff0cClaude 3 Opus\u6a21\u578b\u7684\u51c6\u786e\u7387\u4ece\u96f6\u6837\u672c\u768444.0%\u4e0b\u964d\u5230\u601d\u7ef4\u94feCoT\u768429.6%\u3002\u8fd9\u8868\u660e\uff0c\u601d\u7ef4\u94feCoT\u5f15\u5165\u4e86\u8bed\u8a00\u7684\u5e72\u9884\uff0c\u53ef\u80fd\u4f7f\u6a21\u578b\u66f4\u503e\u5411\u4e8e\u901a\u8fc7\u63cf\u8ff0\u6027\u8bed\u8a00\u6765\u7406\u89e3\u89c6\u89c9\u7279\u5f81\uff0c\u5bfc\u81f4\u5bf9\u9762\u90e8\u7ec6\u8282\u7684\u6355\u6349\u4e0d\u8db3\u3002\u8fd9\u4e0e\u4eba\u7c7b\u5728\u9762\u90e8\u8bc6\u522b\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\u7c7b\u4f3c\uff0c\u5373\u53e3\u5934\u63cf\u8ff0\u4f1a\u5e72\u6270\u89c6\u89c9\u8bb0\u5fc6\u7684\u7ec6\u8282\u5904\u7406\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u4e09\uff1a\u5904\u7406\u5305\u542b\u4f8b\u5916\u7684\u5206\u7c7b<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5728\u5904\u7406\u5305\u542b\u4f8b\u5916\u7684\u5206\u7c7b\u4efb\u52a1\u4e2d\uff0c\u4eba\u7c7b\u901a\u5e38\u503e\u5411\u4e8e\u5bfb\u627e\u7b80\u5355\u3001\u5e7f\u6cdb\u9002\u7528\u7684\u5206\u7c7b\u89c4\u5219\u3002\u7136\u800c\uff0c\u5f53\u6570\u636e\u4e2d\u5305\u542b\u4e0d\u7b26\u5408\u8fd9\u4e9b\u89c4\u5219\u7684\u4f8b\u5916\u65f6\uff0c\u8fd9\u79cd\u7b56\u7565\u4f1a\u5bfc\u81f4\u9519\u8bef\u7684\u5206\u7c7b\u7ed3\u679c\u3002\u7814\u7a76\u8868\u660e\uff0c\u8981\u6c42\u4eba\u7c7b\u5bf9\u5206\u7c7b\u8fc7\u7a0b\u8fdb\u884c\u89e3\u91ca\u65f6\uff0c\u4ed6\u4eec\u66f4\u5bb9\u6613\u9677\u5165\u7b80\u5355\u89c4\u5219\u7684\u9677\u9631\uff0c\u4ece\u800c\u589e\u52a0\u5b66\u4e60\u65f6\u95f4\u548c\u9519\u8bef\u7387\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">2. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5728\u8be5\u5b9e\u9a8c\u4e2d\uff0c\u4f5c\u8005\u6784\u9020\u4e86\u4e00\u7ec4\u8f66\u8f86\u5206\u7c7b\u4efb\u52a1\uff0c\u8981\u6c42\u6a21\u578b\u5bf9\u5177\u6709\u591a\u4e2a\u7279\u5f81\uff08\u5982\u989c\u8272\u3001\u5f62\u72b6\u7b49\uff09\u7684\u8f66\u8f86\u8fdb\u884c\u5206\u7c7b\u3002\u6570\u636e\u96c6\u4e2d\u5305\u542b\u4e00\u4e9b\u4f8b\u5916\u60c5\u51b5\uff0c\u4f7f\u5f97\u5355\u4e00\u7684\u7279\u5f81\u65e0\u6cd5\u5b8c\u5168\u63cf\u8ff0\u5206\u7c7b\u89c4\u5219\u3002\u6a21\u578b\u9700\u8981\u5728\u591a\u8f6e\u8bad\u7ec3\u4e2d\u4e0d\u65ad\u8c03\u6574\u5206\u7c7b\u7b56\u7565\uff0c\u76f4\u5230\u6240\u6709\u8f66\u8f86\u7684\u5206\u7c7b\u5747\u6b63\u786e\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u7ed3\u679c\u663e\u793a\uff0c\u601d\u7ef4\u94feCoT\u663e\u8457\u589e\u52a0\u4e86\u6a21\u578b\u5b66\u4e60\u6b63\u786e\u5206\u7c7b\u7684\u65f6\u95f4\u3002\u4f8b\u5982\uff0cGPT-4o\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u9700\u8981\u7684\u8fed\u4ee3\u6b21\u6570\u662f\u76f4\u63a5\u5206\u7c7b\u7684\u56db\u500d\u4ee5\u4e0a\u3002\u8fd9\u8868\u660e\uff0c\u601d\u7ef4\u94feCoT\u8ba9\u6a21\u578b\u503e\u5411\u4e8e\u5c1d\u8bd5\u627e\u5230\u4e00\u79cd\u666e\u9002\u7684\u89c4\u5219\u6765\u63cf\u8ff0\u6570\u636e\uff0c\u4f46\u7531\u4e8e\u5b58\u5728\u4f8b\u5916\uff0c\u8fd9\u79cd\u7b56\u7565\u53cd\u800c\u589e\u52a0\u4e86\u5b66\u4e60\u7684\u590d\u6742\u5ea6\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u56db\uff1a\u903b\u8f91\u4e0d\u4e00\u81f4\u6027\u68c0\u6d4b<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5728\u67d0\u4e9b\u903b\u8f91\u63a8\u7406\u4efb\u52a1\u4e2d\uff0c\u4eba\u7c7b\u5728\u89e3\u91ca\u903b\u8f91\u4e0d\u4e00\u81f4\u6027\u65f6\uff0c\u5f80\u5f80\u4f1a\u66f4\u503e\u5411\u4e8e\u5bfb\u627e\u4e00\u79cd\u5408\u7406\u7684\u89e3\u91ca\uff0c\u800c\u5ffd\u7565\u4e0d\u4e00\u81f4\u6027\u672c\u8eab\uff0c\u5bfc\u81f4\u5224\u65ad\u9519\u8bef\u3002\u8fd9\u79cd\u73b0\u8c61\u662f\u5426\u4f1a\u5728\u8bed\u8a00\u6a21\u578b\u4e2d\u590d\u73b0\u662f\u4e00\u4e2a\u6709\u8da3\u7684\u95ee\u9898\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">4. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u4f5c\u8005\u5229\u7528\u81ea\u7136\u8bed\u8a00\u63a8\u7406\u6570\u636e\u96c6\uff08\u5982SNLI\u3001MNLI\uff09\u6784\u5efa\u4e86\u903b\u8f91\u4e0d\u4e00\u81f4\u6027\u68c0\u6d4b\u4efb\u52a1\uff0c\u8981\u6c42\u6a21\u578b\u5224\u65ad\u4e24\u53e5\u5b50\u662f\u5426\u5b58\u5728\u903b\u8f91\u4e0d\u4e00\u81f4\u3002\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\uff0c\u6a21\u578b\u88ab\u8981\u6c42\u9996\u5148\u5bf9\u4e24\u53e5\u5b50\u4e4b\u95f4\u7684\u5173\u7cfb\u8fdb\u884c\u89e3\u91ca\uff0c\u7136\u540e\u518d\u4f5c\u51fa\u5224\u65ad\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">5. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u5728\u903b\u8f91\u63a8\u7406\u4efb\u52a1\u4e2d\uff0c\u601d\u7ef4\u94feCoT\u53cd\u800c\u63d0\u9ad8\u4e86\u6a21\u578b\u7684\u8868\u73b0\u3002\u4f8b\u5982\uff0cGPT-4o\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u7684\u51c6\u786e\u7387\u6bd4\u96f6\u6837\u672c\u6761\u4ef6\u63d0\u9ad8\u4e8640%\u4ee5\u4e0a\u3002\u8fd9\u53ef\u80fd\u662f\u56e0\u4e3a\u903b\u8f91\u63a8\u7406\u4efb\u52a1\u9700\u8981\u66f4\u6df1\u5c42\u6b21\u7684\u7406\u89e3\u548c\u63a8\u7406\uff0c\u800c\u601d\u7ef4\u94feCoT\u6b63\u597d\u6709\u52a9\u4e8e\u8fd9\u4e00\u8fc7\u7a0b\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u4e94\uff1a\u7a7a\u95f4\u76f4\u89c9\u4efb\u52a1<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u4eba\u7c7b\u5728\u8fdb\u884c\u7a7a\u95f4\u63a8\u7406\u65f6\uff0c\u5f80\u5f80\u4f9d\u8d56\u4e8e\u89c6\u89c9\u6216\u8fd0\u52a8\u6a21\u62df\uff0c\u800c\u4e0d\u662f\u8bed\u8a00\u63cf\u8ff0\u3002\u4f8b\u5982\uff0c\u503e\u659c\u676f\u5b50\u4f7f\u6c34\u9762\u4fdd\u6301\u6c34\u5e73\u7684\u4efb\u52a1\u4e2d\uff0c\u89c6\u89c9\u6216\u8fd0\u52a8\u76f4\u89c9\u6bd4\u8bed\u8a00\u63cf\u8ff0\u66f4\u4e3a\u6709\u6548\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">2. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u4f5c\u8005\u6a21\u62df\u4e86\u503e\u659c\u676f\u5b50\u7684\u4efb\u52a1\uff0c\u8ba9\u6a21\u578b\u5224\u65ad\u5982\u4f55\u8ba9\u4e24\u4e2a\u4e0d\u540c\u9ad8\u5ea6\u548c\u5f62\u72b6\u7684\u676f\u5b50\u4e2d\u6c34\u9762\u4fdd\u6301\u6c34\u5e73\u3002\u6a21\u578b\u88ab\u63d0\u4f9b\u4e00\u5f20\u5305\u542b\u4e24\u4e2a\u676f\u5b50\u7684\u56fe\u50cf\uff0c\u5e76\u88ab\u8981\u6c42\u9009\u62e9\u6b63\u786e\u7684\u6c34\u4f4d\u9ad8\u5ea6\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u7ed3\u679c\u663e\u793a\uff0c\u601d\u7ef4\u94feCoT\u5e76\u672a\u663e\u8457\u6539\u53d8\u6a21\u578b\u7684\u8868\u73b0\u3002\u7531\u4e8e\u5927\u578b\u591a\u6a21\u6001\u6a21\u578b\u7f3a\u4e4f\u4eba\u7c7b\u7684\u8fd0\u52a8\u6a21\u62df\u80fd\u529b\uff0c\u5176\u63a8\u7406\u4e3b\u8981\u4f9d\u8d56\u4e8e\u8bed\u8a00\u548c\u89c6\u89c9\u8f93\u5165\uff0c\u56e0\u6b64\u5728\u6b64\u7c7b\u4efb\u52a1\u4e2d\u65e0\u6cd5\u83b7\u5f97\u4e0e\u4eba\u7c7b\u7c7b\u4f3c\u7684\u76f4\u89c9\u4f18\u52bf\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u5b9e\u9a8c\u516d\uff1a\u805a\u5408\u591a\u7ef4\u7279\u5f81\u7684\u51b3\u7b56\u4efb\u52a1<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">1. \u5b9e\u9a8c\u80cc\u666f<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u5728\u9700\u8981\u8003\u8651\u591a\u4e2a\u7ef4\u5ea6\u7279\u5f81\u7684\u590d\u6742\u51b3\u7b56\u4efb\u52a1\u4e2d\uff0c\u4f8b\u5982\u9009\u62e9\u6700\u4f73\u516c\u5bd3\uff0c\u4eba\u7c7b\u5f80\u5f80\u4f1a\u56e0\u4fe1\u606f\u91cf\u8fc7\u5927\u800c\u5728\u6df1\u601d\u719f\u8651\u540e\u4f5c\u51fa\u9519\u8bef\u51b3\u7b56\u3002\u800c\u6a21\u578b\u53ef\u4ee5\u901a\u8fc7\u601d\u7ef4\u94feCoT\u6709\u6548\u805a\u5408\u591a\u7ef4\u4fe1\u606f\uff0c\u56e0\u6b64\u8868\u73b0\u53ef\u80fd\u4e0d\u540c\u4e8e\u4eba\u7c7b\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">2. \u5b9e\u9a8c\u8bbe\u8ba1\u4e0e\u6a21\u578b\u9009\u62e9<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u4f5c\u8005\u6784\u9020\u4e86\u4e00\u7ec4\u5305\u542b\u591a\u4e2a\u7279\u5f81\u63cf\u8ff0\u7684\u516c\u5bd3\u9009\u62e9\u4efb\u52a1\uff0c\u8981\u6c42\u6a21\u578b\u6839\u636e\u8fd9\u4e9b\u63cf\u8ff0\u9009\u51fa\u6700\u4f73\u516c\u5bd3\u3002\u7279\u5f81\u63cf\u8ff0\u6570\u91cf\u8fdc\u8d85\u4eba\u7c7b\u7684\u5de5\u4f5c\u8bb0\u5fc6\u5bb9\u91cf\uff0c\u56e0\u6b64\u4efb\u52a1\u5bf9\u6a21\u578b\u548c\u4eba\u7c7b\u7684\u96be\u5ea6\u5dee\u5f02\u8f83\u5927\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3. \u5b9e\u9a8c\u7ed3\u679c\u4e0e\u5206\u6790<\/h5>\n\n\n\n<p class=\"wp-block-paragraph\">\u7ed3\u679c\u663e\u793a\uff0c\u601d\u7ef4\u94feCoT\u5e2e\u52a9\u6a21\u578b\u66f4\u597d\u5730\u805a\u5408\u591a\u7ef4\u4fe1\u606f\uff0c\u4ece\u800c\u63d0\u9ad8\u4e86\u51b3\u7b56\u51c6\u786e\u7387\u3002\u4f8b\u5982\uff0cClaude 3.5 Sonnet\u5728\u601d\u7ef4\u94feCoT\u6761\u4ef6\u4e0b\u7684\u51c6\u786e\u7387\u4ece62%\u63d0\u9ad8\u523072%\u3002\u8fd9\u8868\u660e\uff0c\u6a21\u578b\u7684\u4e0a\u4e0b\u6587\u7a97\u53e3\u80fd\u529b\u4f7f\u5176\u80fd\u591f\u5728\u9762\u5bf9\u590d\u6742\u7684\u591a\u7ef4\u4fe1\u606f\u65f6\uff0c\u501f\u52a9\u601d\u7ef4\u94feCoT\u8fdb\u884c\u66f4\u5168\u9762\u7684\u5206\u6790\u548c\u51b3\u7b56\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u7814\u7a76\u7ed3\u8bba\u4e0e\u5e7f\u6cdb\u5f71\u54cd<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">\u8bba\u6587\u901a\u8fc7\u516d\u7ec4\u5b9e\u9a8c\uff0c\u7cfb\u7edf\u6027\u5730\u5206\u6790\u4e86\u601d\u7ef4\u94feCoT\u5728\u4e0d\u540c\u4efb\u52a1\u4e2d\u7684\u6548\u679c\uff0c\u53d1\u73b0\u5176\u6027\u80fd\u8868\u73b0\u53d6\u51b3\u4e8e\u4efb\u52a1\u7684\u6027\u8d28\u4ee5\u53ca\u6a21\u578b\u4e0e\u4eba\u7c7b\u5728\u4efb\u52a1\u4e0a\u7684\u8ba4\u77e5\u5dee\u5f02\u3002\u5bf9\u4e8e\u90a3\u4e9b\u201c\u6df1\u601d\u719f\u8651\u201d\u4f1a\u635f\u5bb3\u4eba\u7c7b\u8868\u73b0\u7684\u4efb\u52a1\uff0c\u601d\u7ef4\u94feCoT\u901a\u5e38\u4e5f\u4f1a\u5bfc\u81f4\u6a21\u578b\u6027\u80fd\u4e0b\u964d\u3002\u800c\u5728\u67d0\u4e9b\u4efb\u52a1\u4e2d\uff0c\u6a21\u578b\u7531\u4e8e\u5177\u5907\u4e0d\u540c\u4e8e\u4eba\u7c7b\u7684\u8ba4\u77e5\u7279\u70b9\uff0c\u5982\u66f4\u5927\u7684\u4e0a\u4e0b\u6587\u7a97\u53e3\u6216\u5bf9\u903b\u8f91\u63a8\u7406\u7684\u9002\u5e94\u6027\uff0c\u601d\u7ef4\u94feCoT\u53cd\u800c\u5bf9\u5176\u8868\u73b0\u6709\u79ef\u6781\u5f71\u54cd\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8fd9\u79cd\u7814\u7a76\u4e3a\u6211\u4eec\u7406\u89e3\u63d0\u793a\u9009\u62e9\u548c\u63a8\u7406\u65b9\u5f0f\u63d0\u4f9b\u4e86\u65b0\u7684\u89c6\u89d2\uff0c\u63d0\u9192\u6211\u4eec\u4e0d\u80fd\u7b80\u5355\u5730\u5c06\u601d\u7ef4\u94feCoT\u4f5c\u4e3a\u63d0\u5347\u6a21\u578b\u6027\u80fd\u7684\u9ed8\u8ba4\u7b56\u7565\u3002\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\uff0c\u9700\u8981\u7ed3\u5408\u4efb\u52a1\u7279\u5f81\uff0c\u6743\u8861\u662f\u5426\u4f7f\u7528\u601d\u7ef4\u94feCoT\uff0c\u4ee5\u907f\u514d\u4e0d\u5fc5\u8981\u7684\u6027\u80fd\u635f\u5931\u3002\u8fd9\u5bf9\u672a\u6765\u7684\u6a21\u578b\u8bbe\u8ba1\u548c\u63a8\u7406\u7b56\u7565\u4f18\u5316\u5177\u6709\u91cd\u8981\u610f\u4e49\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8bba\u6587Mind Your Step (by Step): Chain-of-Thought can Reduce [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8],"tags":[39,90,86],"class_list":["post-5067","post","type-post","status-publish","format-standard","hentry","category-tech","tag-ai","tag-multimodal","tag-llm"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a>","rttpg_excerpt":"\u8bba\u6587Mind Your Step (by Step): Chain-of-Thought can Reduce&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5067","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5067"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5067\/revisions"}],"predecessor-version":[{"id":5069,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5067\/revisions\/5069"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5067"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5067"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5067"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}