{"id":26152,"date":"2025-05-30T11:03:40","date_gmt":"2025-05-30T03:03:40","guid":{"rendered":"https:\/\/aif.amtbbs.org\/?p=26152"},"modified":"2025-05-30T11:03:40","modified_gmt":"2025-05-30T03:03:40","slug":"ai%e4%bb%85%e5%87%ad%e8%87%aa%e4%bf%a1%e5%ad%a6%e4%bc%9a%e6%8e%a8%e7%90%86%ef%bc%8c%e6%b5%99%e5%a4%a7%e6%a0%a1%e5%8f%8b%e5%a4%8d%e5%88%bbdeepseek%e9%95%bf%e6%80%9d%e7%bb%b4%e9%93%be","status":"publish","type":"post","link":"https:\/\/aif.amtbbs.org\/index.php\/2025\/05\/30\/26152\/","title":{"rendered":"AI\u4ec5\u51ed\u201c\u81ea\u4fe1\u201d\u5b66\u4f1a\u63a8\u7406\uff0c\u6d59\u5927\u6821\u53cb\u590d\u523bDeepSeek\u957f\u601d\u7ef4\u94fe\u6d8c\u73b0\uff0c\u5f3a\u5316\u5b66\u4e60\u65e0\u9700\u5916\u90e8\u5956\u52b1\u4fe1\u53f7"},"content":{"rendered":"<div><img data-dominant-color=\"274460\" data-has-transparency=\"false\" style=\"--dominant-color: #274460;\" loading=\"lazy\" decoding=\"async\" class=\"not-transparent alignnone size-full wp-image-26154\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2025\/05\/14547874703c7425d9d9214b24f72ac7355bab-300x167-1.jpg\" width=\"300\" height=\"167\" alt=\"\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2025\/05\/14547874703c7425d9d9214b24f72ac7355bab-300x167-1.jpg 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2025\/05\/14547874703c7425d9d9214b24f72ac7355bab-300x167-1-150x84.jpg 150w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/div>\n<div><\/div>\n<div class=\"article-desc\">\u5927\u6a21\u578b\u65e0\u9700\u63a5\u89e6\u771f\u5b9e\u7b54\u6848\uff0c\u4ec5\u901a\u8fc7\u4f18\u5316\u81ea\u5df1\u7684\u4fe1\u5fc3\uff0c\u5c31\u80fd\u5b66\u4f1a\u590d\u6742\u63a8\u7406\u3002<\/div>\n<div id=\"postspictures\" class=\"article-content\">\n<div id=\"container\" class=\"container am-engine\" data-v-1d7a5742=\"\" data-element=\"root\">\n<p>\u590d\u523bDeepSeek-R1\u7684\u957f\u601d\u7ef4\u94fe\u63a8\u7406\uff0c\u5927\u6a21\u578b\u5f3a\u5316\u5b66\u4e60\u65b0\u8303\u5f0fRLIF\u6210\u70ed\u95e8\u8bdd\u9898\u3002<\/p>\n<p class=\"js_darkmode__2\">UC Berkeley\u56e2\u961f\u5171\u540c\u4e00\u4f5cXuandong Zhao\u628a\u8fd9\u9879\u6210\u679c\u79f0\u4e3a\uff1a<\/p>\n<blockquote>\n<p class=\"js_darkmode__4\">\u5927\u6a21\u578b\u65e0\u9700\u63a5\u89e6\u771f\u5b9e\u7b54\u6848\uff0c\u4ec5\u901a\u8fc7\u4f18\u5316\u81ea\u5df1\u7684\u4fe1\u5fc3\uff0c\u5c31\u80fd\u5b66\u4f1a\u590d\u6742\u63a8\u7406\u3002<\/p>\n<\/blockquote>\n<p class=\"js_darkmode__5\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202505\/29\/748f253707b28cb5e72900563e5c5a1c79154d.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__6\">\u5177\u4f53\u6765\u8bf4\uff0c\u65b0\u65b9\u6cd5\u5b8c\u5168\u4e0d\u9700\u8981\u5916\u90e8\u5956\u52b1\u4fe1\u53f7\u6216\u6807\u6ce8\u6570\u636e\uff0c\u53ea\u9700\u4f7f\u7528\u6a21\u578b\u81ea\u8eab\u7684\u7f6e\u4fe1\u7a0b\u5ea6\u4f5c\u4e3a\u5185\u5728\u5956\u52b1\u4fe1\u53f7\u3002<\/p>\n<p class=\"js_darkmode__7\">\u4e0e\u4f7f\u7528\u5916\u90e8\u5956\u52b1\u4fe1\u53f7GRPO\u76f8\u6bd4\uff0c\u65b0\u65b9\u6cd5\u5728\u6570\u5b66\u4efb\u52a1\u4e0a\u4e0d\u9700\u8981\u6807\u51c6\u7b54\u6848\u4e5f\u80fd\u63d0\u5347\u57fa\u7840\u6a21\u578b\u6027\u80fd\uff0c\u5728\u4ee3\u7801\u4efb\u52a1\u4e0a\u8868\u73b0\u5f97\u66f4\u597d\u3002<\/p>\n<p class=\"js_darkmode__8\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202505\/29\/e5c078138a2a7e7cc8c780367eff1999f86908.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__9\">\u51e0\u4e4e\u540c\u4e00\u65f6\u95f4\uff0c\u53e6\u5916\u4e00\u7bc7\u8bba\u6587\u300aRENT: Reinforcement Learning via Entropy Minimization\u300b\u4e5f\u9a8c\u8bc1\u4e86\u76f8\u4f3c\u7684\u7ed3\u8bba\u3002<\/p>\n<p class=\"js_darkmode__10\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202505\/29\/a3bf84698d809bf034189192b92e84c2020ca2.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__11\">\u4f5c\u8005\u8868\u793a\u4e24\u8005\u7684\u4e3b\u8981\u533a\u522b\u5728\u4e8e\u4f7f\u7528KL\u6563\u5ea6\u548c\u6700\u5c0f\u5316\u71b5\u8861\u91cf\u81ea\u4fe1\u7a0b\u5ea6\u3002<\/p>\n<p class=\"js_darkmode__12\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202505\/29\/8222bb4917d9b6ec5b30169b75009f5f0a7883.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__13\">Dropbox\u5de5\u7a0b\u526f\u603b\u88c1\u770b\u540e\u8868\u793a\uff1a<strong>Confidence is all you need<\/strong>\u3002<\/p>\n<p class=\"js_darkmode__14\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s4.51cto.com\/oss\/202505\/29\/03b261c104e1c7c93b2370ccac03962d440ce9.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<h2>\u201c\u81ea\u4fe1\u201d\u9a71\u52a8\u7684\u5f3a\u5316\u5b66\u4e60<\/h2>\n<p class=\"js_darkmode__16\">\u957f\u671f\u4ee5\u6765\uff0c\u8bad\u7ec3\u5927\u6a21\u578b\u4e3b\u8981\u4f9d\u8d56\u4e24\u79cd\u65b9\u5f0f\uff1a<\/p>\n<p class=\"js_darkmode__17\">\u8981\u4e48\u9700\u8981\u5927\u91cf\u4eba\u5de5\u6807\u6ce8\uff08\u5982ChatGPT\u7684RLHF\uff09\uff0c\u8981\u4e48\u9700\u8981\u53ef\u9a8c\u8bc1\u7684\u6807\u51c6\u7b54\u6848\uff08\u5982DeepSeek\u7684RLVR\uff09\u3002<\/p>\n<p class=\"js_darkmode__18\">\u524d\u8005\u6210\u672c\u9ad8\u6602\u4e14\u53ef\u80fd\u5f15\u5165\u504f\u89c1\uff0c\u540e\u8005\u5219\u5c40\u9650\u4e8e\u6570\u5b66\u3001\u7f16\u7a0b\u7b49\u6709\u660e\u786e\u7b54\u6848\u7684\u9886\u57df\u3002<\/p>\n<p class=\"js_darkmode__19\">\u90a3\u4e48\u5f53AI\u80fd\u529b\u9010\u6e10\u63a5\u8fd1\u751a\u81f3\u8d85\u8d8a\u4eba\u7c7b\u65f6\uff0c\u80fd\u5426\u8ba9\u6a21\u578b\u4ec5\u51ed\u81ea\u8eab\u4ea7\u751f\u7684\u5185\u5728\u4fe1\u53f7\uff0c\u6446\u8131\u5bf9\u5916\u90e8\u76d1\u7763\u7684\u4f9d\u8d56\uff1f<\/p>\n<p class=\"js_darkmode__20\">\u9488\u5bf9\u8fd9\u4e2a\u95ee\u9898\uff0cUC Berkeley\u56e2\u961f\u63d0\u51fa\u65b0\u8bad\u7ec3\u65b9\u6cd5Intuitor\uff0c\u8ba1\u7b97\u6a21\u578b\u9884\u6d4b\u5206\u5e03\u4e0e\u5747\u5300\u5206\u5e03\u4e4b\u95f4\u7684KL\u6563\u5ea6\u4f5c\u4e3a\u201c\u81ea\u4fe1\u7a0b\u5ea6\u201d\u3002<\/p>\n<p class=\"js_darkmode__21\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202505\/29\/a43f7bd6555a5ee113475171e483ccc725ab9c.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__22\">\u76f8\u5f53\u4e8e\u4eba\u7c7b\u505a\u9898\u65f6\uff0c\u5982\u679c\u5bf9\u7b54\u6848\u6709\u628a\u63e1\u601d\u8def\u4e5f\u4f1a\u66f4\u6e05\u6670\uff0c\u5f53\u81ea\u4fe1\u4e0d\u8db3\u7684\u65f6\u5019\u5f80\u5f80\u9700\u8981\u91cd\u65b0\u601d\u8003\u3002<\/p>\n<p class=\"js_darkmode__23\">\u901a\u8fc7\u4f18\u5316\u8fd9\u4e2a\u5185\u5728\u4fe1\u53f7\uff0cINTUITOR\u9f13\u52b1\u6a21\u578b\u751f\u6210\u5b83\u81ea\u5df1\u201d\u66f4\u6709\u628a\u63e1\u201d\u7684\u56de\u7b54\uff0c\u4e5f\u80fd\u4fc3\u4f7f\u6a21\u578b\u751f\u6210\u66f4\u7ed3\u6784\u5316\u7684\u63a8\u7406\u8fc7\u7a0b\u3002<\/p>\n<p class=\"js_darkmode__24\">\u5728\u5b9e\u9a8c\u4e2d\uff0c1.5B\u548c3B\u7684\u5c0f\u6a21\u578b\u4e5f\u6d8c\u73b0\u51fa\u4e0eDeepSeek-R1\u7c7b\u4f3c\u7684\u957f\u601d\u7ef4\u94fe\u63a8\u7406\u884c\u4e3a\u3002<\/p>\n<p class=\"js_darkmode__25\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202505\/29\/39ad82281a10fa9bc26929dbbc5b3f17c9f1a9.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__26\">\u8bba\u6587\u8fd8\u6307\u51fa\uff0c\u5185\u5728\u5956\u52b1\u4fe1\u53f7\u8fd8\u83b7\u5f97\u4e00\u4e2a\u989d\u5916\u7684\u597d\u5904\uff1a\u4ece\u673a\u5236\u4e0a\u964d\u4f4e\u4e86\u201c\u5956\u52b1\u9ed1\u5ba2\u201d\u7684\u98ce\u9669\u3002<\/p>\n<p class=\"js_darkmode__27\">\u4f20\u7edf\u5916\u90e8\u5956\u52b1\u4fe1\u53f7\u7684\u5f3a\u5316\u5b66\u4e60\u5bb9\u6613\u88ab\u201c\u94bb\u7a7a\u5b50\u201d\uff0c\u5982\u6a21\u578b\u53ef\u80fd\u751f\u6210\u8bed\u6cd5\u6b63\u786e\u4f46\u903b\u8f91\u9519\u8bef\u7684\u4ee3\u7801\u6765\u5339\u914d\u6d4b\u8bd5\u7528\u4f8b\uff0c\u6216\u5728\u6570\u5b66\u9898\u4e2d\u76f4\u63a5\u80cc\u7b54\u6848\u800c\u975e\u63a8\u7406\u3002<\/p>\n<p class=\"js_darkmode__28\">\u5728INTUITOR\u4e2d\uff0c\u56e2\u961f\u53d1\u73b0\u5982\u679c\u4f7f\u7528\u79bb\u7ebf\u5b66\u4e60\uff0c\u5728\u8bad\u7ec3\u7ea6100\u6b65\u7684\u65f6\u5019\u6a21\u578b\u4e5f\u5b66\u4f1a\u4e86\u4f5c\u5f0a\uff1a\u5728\u56de\u7b54\u4e2d\u9644\u52a0\u4e00\u4e2a\u5df2\u7ecf\u89e3\u51b3\u7684\u7b80\u5355\u95ee\u9898\u6765\u63d0\u9ad8\u81ea\u4fe1\u5ea6\u5206\u6570\u3002<\/p>\n<p class=\"js_darkmode__29\">\u4f46\u4f7f\u7528\u5728\u7ebf\u5b66\u4e60\u5c31\u53ef\u4ee5\u907f\u514d\u8fd9\u4e2a\u95ee\u9898\uff0c\u8bc4\u4f30\u6807\u51c6\u968f\u7740\u6a21\u578b\u80fd\u529b\u54e6\u4e00\u8d77\u8fdb\u5316\uff0c\u4f5c\u5f0a\u7b56\u7565\u53d8\u5f97\u65e0\u6548\u3002<\/p>\n<p class=\"js_darkmode__30\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202505\/29\/39389004717be098bb1066c0f99e9d2fcb4df2.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<h2>\u5b9e\u9a8c\u7ed3\u679c\uff1a\u4e0d\u4ec5\u4f1a\u505a\u9898\uff0c\u8fd8\u4f1a\u4e3e\u4e00\u53cd\u4e09<\/h2>\n<p class=\"js_darkmode__32\">\u56e2\u961f\u9996\u5148\u5b9e\u8bc1\u7814\u7a76\u4e86INTUITOR\u6846\u67b6\u5bf9LLMs\u6570\u5b66\u63a8\u7406\u80fd\u529b\u7684\u63d0\u5347\u3002<\/p>\n<p class=\"js_darkmode__33\">\u5b9e\u9a8c\u9009\u53d6<strong>Qwen2.5-1.5B\/3B<\/strong>\u4f5c\u4e3a\u57fa\u7840\u6a21\u578b\uff0c\u4f7f\u7528\u81ea\u6211\u786e\u5b9a\u5ea6\u4f5c\u4e3a\u552f\u4e00\u7684\u5956\u52b1\u4fe1\u53f7\uff0c\u5e76\u5c06\u5176\u5206\u522b\u7f6e\u4e8e<strong>INTUITOR<\/strong>\u548c\u4e24\u4e2a\u57fa\u7ebf\u65b9\u6cd5\uff08GRPO\u3001GRPO-PV\uff09\u5728MATH\u6570\u636e\u96c6\u7684\u9884\u8bad\u7ec3\u4e2d\u3002<\/p>\n<p class=\"js_darkmode__34\">\u4f7f\u7528\u5bf9\u8bdd\u63d0\u793a\uff0c\u6bcf\u6b21\u5904\u7406128\u9053\u9898\u76ee\u5e76\u5404\u751f\u62107\u4e2a\u5019\u9009\u89e3\u51b3\u65b9\u6848\uff0cKL\u60e9\u7f5a\u7cfb\u6570\u8bbe\u7f6e\u4e3a0.005\u3002<\/p>\n<p class=\"js_darkmode__35\">\u5728\u6570\u5b66\u63a8\u7406\u3001\u4ee3\u7801\u751f\u6210\u3001\u6307\u4ee4\u9075\u5faa\u7684\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8fdb\u884c\u6027\u80fd\u8bc4\u4f30\uff0c\u7ed3\u679c\u5982\u56fe\u6240\u793a\uff1a<\/p>\n<p class=\"js_darkmode__36\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s7.51cto.com\/oss\/202505\/29\/e4e65e923186458b492169a8c82b2fda5fdaee.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__37\">\u5b9e\u9a8c\u8868\u660e\uff0c\u5728\u901a\u8fc7INTUITOR\u8fdb\u884c\u5fae\u8c03\u540e\uff0cQwen2.5-1.5B\u4ece\u6700\u521d\u53ea\u4f1a\u8f93\u51fa\u91cd\u590d\u7684\u65e0\u610f\u4e49\u5185\u5bb9\u4e14\u5bf9\u8bdd\u4efb\u52a1\u5f97\u5206\u5747\u4f4e\u4e8e10%\uff0c\u8f6c\u53d8\u4e3a\u65e0\u6548\u8f93\u51fa\u5927\u5e45\u51cf\u5c11\u3001\u54cd\u5e94\u957f\u5ea6\u6709\u6548\u589e\u52a0\u3002<\/p>\n<p class=\"js_darkmode__38\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202505\/29\/8760ea7279a50ca5e5d906682fae30f05158a4.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__39\">\u5728\u7ed3\u6784\u5316\u63a8\u7406\u80fd\u529b\u4e0a\uff0c\u56e2\u961f\u8fd8\u53d1\u73b0<strong>INTUITOR\u65e9\u671f\u5b66\u4e60\u901f\u5ea6\u66f4\u5feb<\/strong>\uff0c\u5982Qwen2.5-3B\u5728GSM8K\u57fa\u51c6\u6d4b\u8bd5\u4e0aINTUITOR\uff080.811\uff09\u59cb\u7ec8\u4f18\u4e8eGRPO\uff080.758\uff09\u3002<\/p>\n<p class=\"js_darkmode__40\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s8.51cto.com\/oss\/202505\/29\/87768b15202e9397ba5853ea5f166e31ecc9af.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__41\">\u6b64\u5916\uff0cINTUITOR\u5728\u591a\u4efb\u52a1\u6cdb\u5316\u4e0a\u4e5f\u8868\u73b0\u4f18\u79c0\uff0c\u4f8b\u5982\u5f53Qwen2.5-3B\u5728\u4ee3\u7801\u751f\u6210\u4efb\u52a1\u4e0a\uff0c\u867d\u7136\u76f8\u5bf9\u6ede\u540e\u4f46\u6301\u7eed\u589e\u957f\uff0c\u6700\u7ec8\u6027\u80fd\u6bd4GRPO\u9ad8<strong>8%<\/strong>\uff0c\u76f8\u5bf9\u63d0\u5347<strong>65%<\/strong>\u3002<\/p>\n<p class=\"js_darkmode__42\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202505\/29\/c6a08e594dd469b953a469cc6e28a6043fd851.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__43\">\u540c\u65f6\u56e2\u961f\u8fd8\u89c2\u5bdf\u5230\uff0c\u5728\u8fdb\u884c\u957f\u94fe\u63a8\u7406\u65f6\uff0cINTUITOR\u6a21\u578b\u5728\u751f\u6210\u5b8c\u6574\u4ee3\u7801\u524d\uff0c\u90fd\u4f1a<strong>\u6dfb\u52a0\u81ea\u7136\u8bed\u8a00\u63a8\u7406<\/strong>\u00a0\uff08\u5982\u201c\u4e3a\u89e3\u51b3X\u95ee\u9898\uff0c\u9700\u5148\u6267\u884cY\u6b65\u9aa4\u201d\uff09\uff0c\u636e\u63a8\u6d4b\u4e5f\u8bb8\u8fd9\u5c31\u662fINTUITOR\u80fd\u591f\u5728\u6d4b\u8bd5\u4e2d\u59cb\u7ec8\u8868\u73b0\u51fa\u8272\u7684\u539f\u56e0\u4e4b\u4e00\u3002<\/p>\n<p class=\"js_darkmode__44\">\u5b83\u7684\u6f14\u8fdb\u8fc7\u7a0b\u5927\u6982\u53ef\u4ee5\u63cf\u8ff0\u4e3a\u4e09\u4e2a\u9636\u6bb5\uff1a<\/p>\n<ol class=\"list-paddingleft-1 js_darkmode__45\" data-id=\"o7c57b9a-DfflWgB1\">\n<li data-id=\"l34939b2-gJkURcab\">\u6a21\u578b\u5b66\u4f1a\u751f\u6210\u4ee3\u7801\uff0c\u5b9e\u73b0\u51c6\u786e\u7387\u63d0\u5347\u548c\u65e0\u6548\u54cd\u5e94\u51cf\u5c11\u3002<\/li>\n<li data-id=\"l34939b2-X1C2hkil\">\u8fdb\u884c\u4ee3\u7801\u524d\u63a8\u7406\u4ee5\u4fc3\u8fdb\u81ea\u6211\u7406\u89e3\u3002<\/li>\n<li data-id=\"l34939b2-HaEdHHAM\">\u9010\u6b65\u7ec6\u5316\u751f\u6210\u5e26\u8be6\u7ec6\u63a8\u7406\u7684\u6709\u6548\u4ee3\u7801\u3002<\/li>\n<\/ol>\n<p class=\"js_darkmode__46\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s9.51cto.com\/oss\/202505\/29\/f58dc6921e063b5c7f35729797f60336d41d28.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__47\">\u4e3a\u4e86\u8bc4\u4f30\u81ea\u6211\u786e\u5b9a\u5ea6\u4f5c\u4e3a\u5956\u52b1\u7684\u9c81\u68d2\u6027\uff0c\u7814\u7a76\u4eba\u5458\u8fd8\u5c06\u79bb\u7ebf\u81ea\u6211\u786e\u5b9a\u5ea6\uff08\u6765\u81ea\u56fa\u5b9a\u57fa\u7840\u6a21\u578b\u7684\u5956\u52b1\uff09\u4e0e\u5728\u7ebf\u81ea\u6211\u786e\u5b9a\u5ea6\uff08\u6765\u81ea\u4e0d\u65ad\u8fdb\u5316\u7684\u7b56\u7565\u6a21\u578b\u7684\u5956\u52b1\uff09\u8fdb\u884c\u4e86\u6bd4\u8f83\u3002<\/p>\n<p class=\"js_darkmode__48\">\u5b9e\u9a8c\u53d1\u73b0\u79bb\u7ebf\u5956\u52b1\u4f1a\u5728100\u6b65\u540e\u901a\u8fc7\u6dfb\u52a0\u65e0\u5173\u5185\u5bb9\u5bfc\u81f4\u51c6\u786e\u7387\u5d29\u6e83\uff0c\u800c<strong>\u5728\u7ebf\u5956\u52b1\u4e0e\u7b56\u7565\u534f\u540c\u6f14\u5316<\/strong>\uff0c\u53ef\u6709\u6548\u9632\u6b62\u7834\u89e3\u3002<\/p>\n<p class=\"js_darkmode__49\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202505\/29\/515c8ad40bbb48eda8a217a16f233c630a2651.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__50\">\u4e3a\u8fdb\u4e00\u6b65\u8bc4\u4f30<strong>\u81ea\u6211\u786e\u5b9a\u5ea6<\/strong>\u4f5c\u4e3a\u5956\u52b1\u4fe1\u53f7\u7684\u8d28\u91cf\uff0c\u7814\u7a76\u4eba\u5458\u8fd8\u5206\u6790\u4e86\u6a21\u578b\u5728MATH500\u54cd\u5e94\u4e2d\u751f\u6210\u7684\u81ea\u6211\u786e\u5b9a\u5ea6\u5206\u6570\u5206\u5e03\u3002<\/p>\n<p class=\"js_darkmode__51\">\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cINTUITOR\u6a21\u578b\u5bf9\u6b63\u786e\u7b54\u6848\u7684self-certainty\u663e\u8457\u66f4\u9ad8\uff0c\u800cGRPO\u867d\u63d0\u5347\u4e86\u6a21\u578b\u81ea\u8bc4\u80fd\u529b\uff0c\u4f46\u533a\u5206\u5ea6\u660e\u663e\u4f4e\u4e8eINTUITOR\u3002<\/p>\n<p class=\"js_darkmode__52\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s8.51cto.com\/oss\/202505\/29\/d29f5b775b4c4c1c3dd86023f029e304a6248c.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__53\">\u7531\u4e8e\u53d7\u8ba1\u7b97\u8d44\u6e90\u9650\u5236\uff0c\u5b9e\u9a8c\u53ea\u5728\u76f8\u5bf9\u8f83\u5c0f\u7684\u65e0\u76d1\u7763\u8bed\u6599\u5e93\u4e0a\u8fdb\u884c\u8bad\u7ec3\uff0c\u672a\u6765\u53ef\u5728\u66f4\u5927\u89c4\u6a21\u7684\u57fa\u7840\u6a21\u578b\u548c\u66f4\u591a\u6837\u5316\u7684\u771f\u5b9e\u4e16\u754c\u6570\u636e\u96c6\u4e0a\u8fdb\u4e00\u6b65\u7814\u7a76INTUITOR\u7684\u4f18\u52bf\u3002<\/p>\n<h2>\u56e2\u961f\u4ecb\u7ecd<\/h2>\n<p class=\"js_darkmode__55\">\u672c\u9879\u7814\u7a76\u6765\u81eaUC Berkeley\u7684Sergey Levine\u3001\u5b8b\u6653\u4e1c\u56e2\u961f\uff0c\u4f5c\u8005\u4e00\u5171\u6709\u4e94\u4f4d\uff0c\u5206\u522b\u662f\u7b2c\u4e00\u4f5c\u8005\u535a\u58eb\u540e\u7814\u7a76\u5458Xuandong Zhao\u3001\u5171\u540c\u4e00\u4f5c\u672c\u79d1\u751fZhewei Kang\u3001\u6765\u81ea\u8036\u9c81\u5927\u5b66\u7684Aosong Feng\uff0c\u4ee5\u53caSergey Levine\u548cDawn Song\u3002<\/p>\n<p class=\"js_darkmode__56\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s8.51cto.com\/oss\/202505\/29\/18e91978303740a2ec6735057df05aec3114c0.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__57\">2019\u5e74\uff0cXuandong Zhao\u4ece\u6d59\u6c5f\u5927\u5b66\u6bd5\u4e1a\u540e\uff0c\u5c31\u8fdb\u5165\u4e86\u52a0\u5dde\u5927\u5b66\u5723\u5854\u82ad\u82ad\u62c9\u5206\u6821\u653b\u8bfb\u8ba1\u7b97\u673a\u79d1\u5b66\u535a\u58eb\u5b66\u4f4d\uff0c\u671f\u95f4\u8fd8\u66fe\u5728\u963f\u91cc\u5df4\u5df4\u3001Microsoft\u548cGoogle\u7b49\u516c\u53f8\u5b9e\u4e60\u3002<\/p>\n<p class=\"js_darkmode__58\">\u81ea2024\u5e74\u4ed6\u8fdb\u5165UC Berkeley\u540e\uff0c\u9664\u672c\u6b21\u7684\u65b0\u6210\u679c\u5916\uff0c\u81f3\u4eca\u4e00\u5171\u8fd8\u53d1\u8868\u8fc7\u5341\u591a\u7bc7\u8bba\u6587\uff0c\u5e76\u5148\u540e\u88abICLR 2025\u3001ICML 2025\u7b49\u63a5\u6536\u3002<\/p>\n<p class=\"js_darkmode__59\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s7.51cto.com\/oss\/202505\/29\/836137908163a882f0e40733cde652eb1f58b7.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__60\">\u53e6\u5916\u5728\u4eca\u5e742\u6708\uff0cXuandong Zhao\u548cZhewei Kang\u8fd8\u5408\u4f5c\u53d1\u8868\u4e86\u4e00\u7bc7\u8bba\u6587\uff0c\u63cf\u8ff0\u4e86<strong>\u57fa\u4e8e\u81ea\u6211\u786e\u5b9a\u6027\u7684LLMs\u63a8\u7406\u80fd\u529b\u63d0\u5347\u65b0\u7b56\u7565Best-of-N<\/strong>\uff0c\u53ef\u4ee5\u770b\u4f5c\u662f\u672c\u7bc7\u8bba\u6587\u7684\u4e00\u6b21\u5148\u9a8c\u5c1d\u8bd5\u3002<\/p>\n<p class=\"js_darkmode__61\"><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202505\/29\/993e30b17da3fb688da176de56bed7199705e6.webp\" alt=\"\u56fe\u7247\" data-type=\"inline\" \/><\/p>\n<p class=\"js_darkmode__62\">\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/abs\/2505.19590<br \/>\n\u4ee3\u7801\u94fe\u63a5\uff1ahttps:\/\/github.com\/sunblaze-ucb\/Intuitor<\/p>\n<p>\u6587\u7ae0\u6765\u81ea\uff1a51CTO<\/p>\n<\/div>\n<\/div>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_26152\" class=\"pvc_stats total_only  \" data-element-id=\"26152\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>\u5927\u6a21\u578b\u65e0\u9700\u63a5\u89e6\u771f\u5b9e\u7b54\u6848\uff0c\u4ec5\u901a\u8fc7\u4f18\u5316\u81ea\u5df1\u7684\u4fe1\u5fc3\uff0c\u5c31\u80fd\u5b66\u4f1a\u590d\u6742\u63a8\u7406\u3002 \u590d\u523bDeepSeek-R1\u7684\u957f\u601d\u7ef4\u94fe\u63a8\u7406\uff0c\u5927 [&hellip;]<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_26152\" class=\"pvc_stats total_only  \" data-element-id=\"26152\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"author":56,"featured_media":26154,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[23,20,80],"tags":[1731],"class_list":["post-26152","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-23","category-20","category-80","tag-aideepseek"],"_links":{"self":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/26152","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/users\/56"}],"replies":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/comments?post=26152"}],"version-history":[{"count":1,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/26152\/revisions"}],"predecessor-version":[{"id":26155,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/26152\/revisions\/26155"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media\/26154"}],"wp:attachment":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media?parent=26152"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/categories?post=26152"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/tags?post=26152"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}