{"id":36087,"date":"2025-01-05T17:00:10","date_gmt":"2025-01-05T09:00:10","guid":{"rendered":"https:\/\/17aitech.com\/?p=36087"},"modified":"2025-01-05T17:00:10","modified_gmt":"2025-01-05T09:00:10","slug":"%e9%a6%96%e4%b8%aao1%e5%a4%8d%e7%8e%b0%e5%bc%80%e6%ba%90rl%e6%a1%86%e6%9e%b6openr%e6%9d%a5%e4%ba%86%ef%bc%8cucl%e3%80%81%e4%b8%8a%e4%ba%a4%e7%ad%89%e9%ab%98%e6%a0%a1%e8%81%94%e5%90%88%e5%9b%a2","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=36087","title":{"rendered":"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03"},"content":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-10-14-5\" target=\"_blank\">\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03<\/a><\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png\"><\/a><\/p>\n<blockquote data-author-name=\"\" data-content-utf8-length=\"166\" data-source-title=\"\" data-type=\"2\" data-url=\"\">\n<section>\n<p>AIxiv\u4e13\u680f\u662f\u673a\u5668\u4e4b\u5fc3\u53d1\u5e03\u5b66\u672f\u3001\u6280\u672f\u5185\u5bb9\u7684\u680f\u76ee\u3002\u8fc7\u53bb\u6570\u5e74\uff0c\u673a\u5668\u4e4b\u5fc3AIxiv\u4e13\u680f\u63a5\u6536\u62a5\u9053\u4e862000\u591a\u7bc7\u5185\u5bb9\uff0c\u8986\u76d6\u5168\u7403\u5404\u5927\u9ad8\u6821\u4e0e\u4f01\u4e1a\u7684\u9876\u7ea7\u5b9e\u9a8c\u5ba4\uff0c\u6709\u6548\u4fc3\u8fdb\u4e86\u5b66\u672f\u4ea4\u6d41\u4e0e\u4f20\u64ad\u3002\u5982\u679c\u60a8\u6709\u4f18\u79c0\u7684\u5de5\u4f5c\u60f3\u8981\u5206\u4eab\uff0c\u6b22\u8fce\u6295\u7a3f\u6216\u8005\u8054\u7cfb\u62a5\u9053\u3002\u6295\u7a3f\u90ae\u7bb1\uff1aliyazhou@jiqizhixin.com\uff1bzhaoyunfeng@jiqizhixin.com<\/p>\n<\/section>\n<\/blockquote>\n<p><strong>OpenR \u7814\u7a76\u56e2\u961f\u6210\u5458\u5305\u62ec\uff1a\u6c6a\u519b\u6559\u6388\uff0c\u4f26\u6566\u5927\u5b66\u5b66\u9662\uff08UCL\uff09\u8ba1\u7b97\u673a\u7cfb\u6559\u6388\uff0c\u963f\u5170\u30fb\u56fe\u7075\u7814\u7a76\u6240 Turing Fellow\uff0c\u5176\u6307\u5bfc\u7684 UCL \u4e00\u5e74\u7ea7\u535a\u58eb\u751f\u5b8b\u7814\u3002\u5229\u7269\u6d66\u5927\u5b66\u52a9\u7406\u6559\u6388\u65b9\u8499\u3002\u4e0a\u6d77\u4ea4\u901a\u5927\u5b66 Apex \u548c\u591a\u667a\u80fd\u4f53\u5b9e\u9a8c\u5ba4\u5f20\u4f1f\u6960\u6559\u6388\uff08\u4e0a\u6d77\u4ea4\u901a\u5927\u5b66\u8ba1\u7b97\u673a\u7cfb\u6559\u6388\u3001\u535a\u58eb\u751f\u5bfc\u5e08\u3001\u526f\u7cfb\u4e3b\u4efb\uff09\uff0c\u6e29\u9896\u526f\u6559\u6388\uff08\u4e0a\u6d77\u4ea4\u901a\u5927\u5b66\u7ea6\u7ff0\u30fb\u970d\u666e\u514b\u7f57<\/strong><strong>\u592b\u7279\u8ba1\u7b97\u673a\u79d1\u5b66\u4e2d\u5fc3\u526f\u6559\u6388\uff09\u4ee5\u53ca\u5176\u6307\u5bfc\u7684\u535a\u58eb\u751f<\/strong><strong>\u4e07\u6893\u715c\u3001\u6e29\u7766\u5b81\u3001\u6731\u5bb6\u741b\u3002\u5f20\u4f1f\u6960\u6559\u6388\u548c\u6e29\u9896\u526f\u6559\u6388<\/strong><strong>\u535a\u58eb\u671f\u95f4\u5c31\u8bfb\u4e8e UCL\uff0c<\/strong><strong>\u6307\u5bfc\u6559\u5e08\u4e3a\u6c6a\u519b\u6559\u6388\u3002\u9999\u6e2f\u79d1\u6280\u5927\u5b66\uff08\u5e7f\u5dde\uff09\u521b\u6821\u6821\u957f\uff0c\u502a\u660e\u9009\uff08Lionel M. Ni\uff09\uff0c\u9999\u6e2f\u5de5\u7a0b<\/strong><strong>\u79d1\u5b66\u9662\u9662\u58eb\uff0c\u9999\u6e2f\u79d1\u6280\u5927\u5b66\uff08\u5e7f\u5dde\uff09\u8bb2\u5e2d\u6559\u6388\u3002\u9648\u96f7\uff0c\u9999\u6e2f\u79d1\u6280\u5927\u5b66\uff08\u5e7f\u5dde\uff09\u4fe1\u606f\u67a2\u7ebd\u9662\u957f\uff0c\u8bb2\u5e2d\u6559\u6388\u3002\u9999\u6e2f\u79d1\u6280\u5927\u5b66\uff08\u5e7f\u5dde\uff09\u4e00\u5e74\u7ea7\u535a\u58eb\u751f\u5218\u5b89\u6770\u3001\u9f9a\u5b50\u94a6\u53d7\u6c6a\u519b\u6559\u6388\u548c\u6768\u6797\u6613\u535a\u58eb\u8054\u5408\u6267\u5bfc\uff0c\u4ee5\u53ca\u897f\u6e56\u5927\u5b66\u5de5\u5b66\u9662\u52a9\u7406\u6559\u6388\uff08\u7814\u7a76\uff09\u6768\u6797\u6613\u3002<\/strong><\/p>\n<section>o1 \u4f5c\u4e3a OpenAI \u5728\u63a8\u7406\u9886\u57df\u7684\u6700\u65b0\u6a21\u578b\uff0c\u5927\u5e45\u5ea6\u63d0\u5347\u4e86 GPT-4o \u5728\u63a8\u7406\u4efb\u52a1\u4e0a\u7684\u8868\u73b0\uff0c\u751a\u81f3\u8d85\u8fc7\u4e86\u5e73\u5747\u4eba\u7c7b\u6c34\u5e73\u3002o1 \u80cc\u540e\u7684\u6280\u672f\u5230\u5e95\u662f\u4ec0\u4e48\uff1fOpenAI \u6280\u672f\u62a5\u544a\u4e2d\u6240\u5f3a\u8c03\u7684\u5f3a\u5316\u5b66\u4e60\u548c\u63a8\u65ad\u9636\u6bb5\u7684 Scaling Law \u5982\u4f55\u5b9e\u73b0\uff1f<\/section>\n<section><\/section>\n<section>\u4e3a\u4e86\u5c1d\u8bd5\u56de\u7b54\u8fd9\u4e9b\u95ee\u9898\uff0c\u4f26\u6566\u5927\u5b66\u5b66\u9662\uff08UCL\uff09\u3001\u4e0a\u6d77\u4ea4\u901a\u5927\u5b66\u3001\u5229\u7269\u6d66\u5927\u5b66\u3001\u9999\u6e2f\u79d1\u6280\u5927\u5b66\uff08\u5e7f\u5dde\uff09\u3001\u897f\u6e56\u5927\u5b66\u8054\u5408\u5f00\u6e90\u4e86\u9996\u4e2a\u7c7b o1 \u5168\u94fe\u6761\u8bad\u7ec3\u6846\u67b6\u300cOpenR\u300d\uff0c\u4e00\u4e2a\u5f00\u6e90\u4ee3\u7801\u5e93\uff0c\u5e2e\u52a9\u7528\u6237\u5feb\u901f\u5b9e\u73b0\u6784\u5efa\u81ea\u5df1\u7684\u590d\u6742\u63a8\u65ad\u6a21\u578b \u3002\u6574\u4e2a\u9879\u76ee\u7531 UCL \u6c6a\u519b\u6559\u6388\u53d1\u8d77\u548c\u6307\u5bfc\uff0c\u5b9e\u9a8c\u4e3b\u8981\u7531\u4e0a\u6d77\u4ea4\u5927\u56e2\u961f\u5b8c\u6210\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-9b9aef8bd05df2215fd463bc5a5e71aa.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-9b9aef8bd05df2215fd463bc5a5e71aa.png\"><\/a><\/p>\n<section>\u6211\u4eec\u4ecb\u7ecd\u4e86 OpenR\uff0c\u9996\u4e2a\u96c6\u8fc7\u7a0b\u5956\u52b1\u6a21\u578b\uff08PRM\uff09\u8bad\u7ec3\u3001\u5f3a\u5316\u5b66\u4e60\u3001\u591a\u79cd\u641c\u7d22\u6846\u67b6\u4e3a\u4e00\u8eab\u7684\u7c7b o1 \u5f00\u6e90\u6846\u67b6\uff0c\u65e8\u5728\u589e\u5f3a\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u7684\u590d\u6742\u63a8\u7406\u80fd\u529b\u3002<\/section>\n<section><\/section>\n<ul>\n<li>\n<section>\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/github.com\/openreasoner\/openr\/blob\/main\/reports\/OpenR-Wang.pdf<\/section>\n<\/li>\n<li>\n<section>\u4ee3\u7801\u94fe\u63a5\uff1ahttps:\/\/github.com\/openreasoner\/openr<\/section>\n<\/li>\n<li>\n<section>\u6559\u7a0b\u94fe\u63a5\uff1ahttps:\/\/openreasoner.github.io\/<\/section>\n<\/li>\n<\/ul>\n<section><\/section>\n<section>OpenR \u5c06\u6570\u636e\u83b7\u53d6\u3001\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\uff08\u5305\u62ec\u5728\u7ebf\u548c\u79bb\u7ebf\u8bad\u7ec3\uff09\u4ee5\u53ca\u975e\u81ea\u56de\u5f52\u89e3\u7801\u96c6\u6210\u5230\u4e00\u4e2a\u7edf\u4e00\u7684\u5e73\u53f0\u4e2d\u3002\u53d7\u5230 OpenAI \u7684 o1 \u6a21\u578b\u6210\u529f\u7684\u542f\u53d1\uff0c OpenR \u91c7\u7528\u4e86\u4e00\u79cd\u57fa\u4e8e\u6a21\u578b\u7684\u65b9\u6cd5\uff0c\u8d85\u8d8a\u4e86\u4f20\u7edf\u7684\u81ea\u56de\u5f52\u65b9\u6cd5\u3002\u6211\u4eec\u901a\u8fc7\u5728 MATH \u6570\u636e\u96c6\u4e0a\u7684\u8bc4\u4f30\u6765\u5c55\u793a OpenR \u7684\u6709\u6548\u6027\uff0c\u5229\u7528\u516c\u5f00\u7684\u6570\u636e\u548c\u641c\u7d22\u65b9\u6cd5\u3002\u521d\u6b65\u5b9e\u9a8c\u8868\u660e\uff0c\u76f8\u5bf9\u6539\u8fdb\u8fbe\u5230\u4e86\u663e\u8457\u63d0\u5347\u3002\u6211\u4eec\u5f00\u6e90\u4e86 OpenR \u6846\u67b6\uff0c\u5305\u62ec\u4ee3\u7801\u3001\u6a21\u578b\u548c\u6570\u636e\u96c6\uff0c\u6211\u4eec\u65e8\u5728\u63a8\u52a8\u63a8\u7406\u9886\u57df\u5f00\u6e90\u793e\u533a\u7684\u53d1\u5c55\uff0c\u6b22\u8fce\u611f\u5174\u8da3\u7684\u4ece\u4e1a\u4eba\u5458\u52a0\u5165\u5230\u6211\u4eec\u7684\u5f00\u6e90\u793e\u533a\u3002\u4ee3\u7801\u3001\u6587\u6863\u3001\u6559\u7a0b\u53ef\u901a\u8fc7 https:\/\/openreasoner.github.io \u8bbf\u95ee\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-4f059eef44c00262207819cf61b5a0d7.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-4f059eef44c00262207819cf61b5a0d7.png\"><\/a><\/p>\n<section><em><sup>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u56fe 1: \u7cfb\u7edf\u8bbe\u8ba1\u56fe<\/sup><\/em><\/section>\n<section><\/section>\n<section><strong>\u7cfb\u7edf\u8bbe\u8ba1\u3002<\/strong>\u8fc7\u7a0b\u5956\u52b1\u6a21\u578b (PRM) \u5728\u4e24\u4e2a\u5173\u952e\u65b9\u9762\u589e\u5f3a\u4e86 LLM \u7684\u7b56\u7565\u3002\u9996\u5148\uff0c\u5728\u8bad\u7ec3\u671f\u95f4\uff0cPRM \u901a\u8fc7\u7b56\u7565\u4f18\u5316\u6280\u672f\uff08\u5982\u4e0a\u56fe\u6240\u793a\u7684\u7b56\u7565\u8fed\u4ee3\uff09\u6539\u8fdb LLM \u7b56\u7565\u3002\u5176\u6b21\uff0c\u5728\u89e3\u7801\u9636\u6bb5\uff0cPRM \u5f15\u5bfc LLM \u7684\u641c\u7d22\u8fc7\u7a0b\uff0c\u4f7f\u63a8\u7406\u671d\u7740\u66f4\u6709\u6548\u7684\u7ed3\u679c\u53d1\u5c55\uff08\u5982\u4e0a\u56fe\u6240\u793a\uff09\u3002\u63a5\u4e0b\u6765\u6211\u4eec\u5c06\u5c55\u793a\uff0cLLM \u7b56\u7565\u8fd8\u53ef\u4ee5\u5e2e\u52a9\u8bc6\u522b\u7f3a\u5931\u7684\u4e2d\u95f4\u63a8\u7406\u6b65\u9aa4\uff0c\u8fd9\u53cd\u8fc7\u6765\u53c8\u53ef\u4ee5\u8fdb\u4e00\u6b65\u8bad\u7ec3\u548c\u6539\u8fdb PRM\u3002\u6b63\u5982\u4e0a\u56fe\u6240\u793a\uff0c\u8fd9\u79cd\u8fed\u4ee3\u7684\u4e92\u52a8\u4f7f LLM \u548c PRM \u80fd\u591f\u6301\u7eed\u5730\u91ca\u653e\u5404\u81ea\u7684\u6f5c\u529b\u4ee5\u6539\u8fdb\u63a8\u7406\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-f30fb93a9e6912931f4a08f7f20f85e8.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-f30fb93a9e6912931f4a08f7f20f85e8.png\"><\/a><\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-cf6432b34f2dc802d8558b6f6fd18e8d.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-cf6432b34f2dc802d8558b6f6fd18e8d.png\"><\/a><\/p>\n<section><em><sup>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0\u56fe 2 \u4ee3\u7801\u7ed3\u6784\u56fe<\/sup><\/em><\/section>\n<section><\/section>\n<section><strong>\u6570\u636e\u589e\u5f3a.<\/strong>\u5728\u4f7f\u7528 LLM \u8fdb\u884c\u63a8\u7406\u65f6\uff0c\u6211\u4eec\u4e0d\u4ec5\u4ec5\u4f9d\u8d56\u6700\u7ec8\u7b54\u6848\u7684\u53cd\u9988\uff0c\u800c\u662f\u4f7f\u7528\u66f4\u8be6\u7ec6\u7684\u53cd\u9988\u65b9\u5f0f\uff0c\u9010\u6b65\u6536\u96c6\u548c\u6807\u6ce8\u6570\u636e\u3002\u8fd9\u6837\u53ef\u4ee5\u5728\u95ee\u9898\u89e3\u51b3\u7684\u8fc7\u7a0b\u4e2d\uff0c\u8bc6\u522b\u51fa\u5177\u4f53\u7684\u9519\u8bef\u4f4d\u7f6e\u5e76\u7ed9\u51fa\u53cd\u9988\uff0c\u4ece\u800c\u5e2e\u52a9\u6a21\u578b\u66f4\u597d\u5730\u5b66\u4e60\u548c\u6539\u8fdb\u3002<\/section>\n<section><\/section>\n<section><strong>MATH-APS.<\/strong>\u6211\u4eec\u901a\u8fc7\u81ea\u52a8\u751f\u6210\u5408\u6210\u6837\u672c\u6765\u589e\u5f3a\u6570\u636e\u3002\u4e0e\u4f9d\u8d56\u6602\u8d35\u4e14\u96be\u4ee5\u6269\u5c55\u7684\u4eba\u5de5\u6807\u6ce8\u7684 PRM800k \u6570\u636e\u96c6\u4e0d\u540c\uff0c\u6211\u4eec\u5f15\u5165\u4e86\u4e00\u4e2a\u65b0\u6570\u636e\u96c6 MATH-APS\u3002\u8fd9\u4e2a\u6570\u636e\u96c6\u57fa\u4e8e MATH \u6570\u636e\u96c6\uff0c\u5e76\u4f7f\u7528 OmegaPRM \u7b49\u81ea\u52a8\u5316\u65b9\u6cd5\u6765\u751f\u6210\u6837\u672c\uff0c\u4ece\u800c\u51cf\u5c11\u4e86\u5bf9\u4eba\u5de5\u6807\u6ce8\u7684\u4f9d\u8d56\uff0c\u66f4\u6613\u4e8e\u5927\u89c4\u6a21\u6536\u96c6\u6570\u636e\u3002\u81ea\u52a8\u5316\u65b9\u6cd5\u5982 OmegaPRM\u3001Math-Shepherd \u548c MiPS \u53ef\u4ee5\u9ad8\u6548\u5730\u6536\u96c6\u9ad8\u8d28\u91cf\u7684\u8fc7\u7a0b\u76d1\u7763\u6570\u636e\u3002\u867d\u7136 Math-Shepherd \u548c MiPS \u63d0\u4f9b\u4e86\u8fc7\u7a0b\u76d1\u7763\u7684\u81ea\u52a8\u5316\u6807\u6ce8\uff0c\u4f46\u5b83\u4eec\u9700\u8981\u5927\u91cf\u7684\u7b56\u7565\u8c03\u7528\uff0c\u8ba1\u7b97\u6210\u672c\u8f83\u9ad8\u3002OmegaPRM \u6539\u8fdb\u4e86\u8fd9\u4e2a\u8fc7\u7a0b\uff0c\u901a\u8fc7\u8fed\u4ee3\u5730\u5212\u5206\u89e3\u51b3\u65b9\u6848\u3001\u56de\u6eaf\u5206\u6790\u5e76\u627e\u51fa\u6a21\u578b\u63a8\u7406\u4e2d\u7684\u7b2c\u4e00\u4e2a\u9519\u8bef\u6b65\u9aa4\u6765\u63d0\u9ad8\u6548\u7387\u3002<\/section>\n<section><\/section>\n<section>\u6211\u4eec\u901a\u8fc7\u81ea\u52a8\u751f\u6210\u5408\u6210\u6837\u672c\u6765\u589e\u5f3a\u6570\u636e\u3002\u4e0e\u4f9d\u8d56\u6602\u8d35\u4e14\u96be\u4ee5\u6269\u5c55\u7684\u4eba\u5de5\u6807\u6ce8\u7684 PRM800k \u6570\u636e\u96c6\u4e0d\u540c\uff0c\u6211\u4eec\u5f15\u5165\u4e86\u4e00\u4e2a\u65b0\u6570\u636e\u96c6 MATH-APS\u3002\u8fd9\u4e2a\u6570\u636e\u96c6\u57fa\u4e8e MATH \u6570\u636e\u96c6\uff0c\u5e76\u4f7f\u7528 OmegaPRM \u7b49\u81ea\u52a8\u5316\u65b9\u6cd5\u6765\u751f\u6210\u6837\u672c\uff0c\u4ece\u800c\u51cf\u5c11\u4e86\u5bf9\u4eba\u5de5\u6807\u6ce8\u7684\u4f9d\u8d56\uff0c\u66f4\u6613\u4e8e\u5927\u89c4\u6a21\u6536\u96c6\u6570\u636e\u3002\u81ea\u52a8\u5316\u65b9\u6cd5\u5982 OmegaPRM\u3001Math-Shepherd \u548c MiPS \u53ef\u4ee5\u9ad8\u6548\u5730\u6536\u96c6\u9ad8\u8d28\u91cf\u7684\u8fc7\u7a0b\u76d1\u7763\u6570\u636e\u3002\u867d\u7136 Math-Shepherd \u548c MiPS \u63d0\u4f9b\u4e86\u8fc7\u7a0b\u76d1\u7763\u7684\u81ea\u52a8\u5316\u6807\u6ce8\uff0c\u4f46\u5b83\u4eec\u9700\u8981\u5927\u91cf\u7684\u7b56\u7565\u8c03\u7528\uff0c\u8ba1\u7b97\u6210\u672c\u8f83\u9ad8\u3002OmegaPRM \u6539\u8fdb\u4e86\u8fd9\u4e2a\u8fc7\u7a0b\uff0c\u901a\u8fc7\u8fed\u4ee3\u5730\u5212\u5206\u89e3\u51b3\u65b9\u6848\u3001\u56de\u6eaf\u5206\u6790\u5e76\u627e\u51fa\u6a21\u578b\u63a8\u7406\u4e2d\u7684\u7b2c\u4e00\u4e2a\u9519\u8bef\u6b65\u9aa4\u6765\u63d0\u9ad8\u6548\u7387\u3002<\/section>\n<section><\/section>\n<section><strong>PRM \u7684\u76d1\u7763\u8bad\u7ec3\u3002<\/strong>\u5728\u8fc7\u7a0b\u5956\u52b1\u6a21\u578b (PRM) \u4e2d\uff0c\u4e3b\u8981\u76ee\u7684\u662f\u5224\u65ad\u89e3\u51b3\u65b9\u6848\u7684\u6b65\u9aa4\u662f\u5426\u5728\u6b63\u786e\u7684\u8f68\u9053\u4e0a\u3002\u56e0\u6b64\uff0cPRM \u4f1a\u8f93\u51fa\u4e00\u4e2a 0 \u5230 1 \u4e4b\u95f4\u7684\u5206\u6570\uff0c\u4f5c\u4e3a\u5f53\u524d\u89e3\u51b3\u8fc7\u7a0b\u7684\u6b63\u786e\u6027\u6307\u6807\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u7ed9\u5b9a\u4e00\u4e2a\u95ee\u9898\u53ca\u5176\u89e3\u51b3\u6b65\u9aa4\u5e8f\u5217\uff0cPRM \u4f1a\u4e3a\u6bcf\u4e00\u6b65\u8ba1\u7b97\u51fa\u4e00\u4e2a\u5206\u6570\uff0c\u8fd9\u53ef\u4ee5\u89c6\u4e3a\u4e00\u4e2a\u4e8c\u5143\u5206\u7c7b\u4efb\u52a1\uff1a\u662f\u5426\u6b63\u786e\u3002\u6211\u4eec\u901a\u8fc7\u5728\u5927\u578b\u8bed\u8a00\u6a21\u578b (LLM) \u4e0a\u7684\u76d1\u7763\u5fae\u8c03\u6765\u8bad\u7ec3 PRM\uff0c\u5c06\u6b63\u786e\u6216\u9519\u8bef\u7684\u5224\u5b9a\u4f5c\u4e3a\u5206\u7c7b\u6807\u7b7e\uff0c\u5e76\u8fdb\u4e00\u6b65\u4f7f\u7528 LLM \u6765\u9884\u6d4b\u6bcf\u4e00\u6b65\u7684\u540e\u7eed\u6807\u8bb0\u3002<\/section>\n<section><\/section>\n<section>Math-psa PRM \u901a\u8fc7\u5728 LLM \u4e0a\u7684\u76d1\u7763\u5fae\u8c03\u6765\u8bad\u7ec3\uff0c\u6b63\u786e \/ \u9519\u8bef\u7684\u533a\u5206\u4f5c\u4e3a\u5206\u7c7b\u6807\u7b7e\u3002\u6211\u4eec\u4f7f\u7528\u6570\u636e\u96c6\u5982 PRM800K\uff0cMath-Shepherd \u4ee5\u53ca\u6211\u4eec\u81ea\u5df1\u7684 MATH-APS \u6570\u636e\u96c6\u6765\u8bad\u7ec3\u4e00\u4e2a\u540d\u4e3a Math-psa \u7684 PRM\u3002\u8fd9\u4e9b\u6570\u636e\u96c6\u7531\u4e09\u4e2a\u90e8\u5206\u7ec4\u6210\uff1a\u95ee\u9898\u3001\u8fc7\u7a0b \u548c \u6807\u7b7e\u3002\u8f93\u5165\u7531 \u95ee\u9898 \u548c \u8fc7\u7a0b \u7684\u62fc\u63a5\u7ec4\u6210\u3002\u5728 \u8fc7\u7a0b \u4e2d\uff0c\u89e3\u51b3\u65b9\u6848\u88ab\u5206\u4e3a\u591a\u4e2a\u6b65\u9aa4\uff0c\u6bcf\u4e2a\u6b65\u9aa4\u7528\u4e00\u4e2a\u7279\u6b8a\u7684\u6b65\u9aa4\u6807\u8bb0\u5206\u9694\uff0c\u4ee5\u6807\u8bb0\u6bcf\u4e2a\u6b65\u9aa4\u7ed3\u675f\u7684\u4f4d\u7f6e\uff0cPRM \u53ef\u4ee5\u5728\u6b64\u5904\u8fdb\u884c\u9884\u6d4b\u3002\u6807\u7b7e\u5bf9\u6574\u4e2a\u8fc7\u7a0b\u8fdb\u884c\u5206\u7c7b\uff0c\u6839\u636e\u89e3\u51b3\u65b9\u6848\u7684\u6b63\u786e\u6027\u5c06\u6bcf\u4e2a\u6b65\u9aa4\u6807\u8bb0\u4e3a + \u6216 -\u3002<\/section>\n<section><\/section>\n<section>\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\uff0c\u6a21\u578b\u4f1a\u5728\u6bcf\u4e2a\u6b65\u9aa4\u6807\u8bb0\u4e4b\u540e\u9884\u6d4b\u6b63\u6216\u8d1f\u6807\u7b7e\u3002\u8f93\u5165\u7684\u62fc\u63a5\u683c\u5f0f\u5305\u542b\u4e86 \u95ee\u9898 \u548c\u5404\u4e2a\u6b65\u9aa4\u4e4b\u95f4\u7684\u6807\u8bb0\u7b26\u3002\u6807\u7b7e\u4ec5\u5206\u914d\u5728\u6b65\u9aa4\u6807\u8bb0\u7b26\u7684\u4f4d\u7f6e\uff0c\u5e76\u5728\u8ba1\u7b97\u635f\u5931\u65f6\u5ffd\u7565\u5176\u4ed6\u4f4d\u7f6e\u3002\u8fd9\u79cd\u65b9\u5f0f\u786e\u4fdd\u6a21\u578b\u8bad\u7ec3\u65f6\u4e3b\u8981\u5173\u6ce8\u8f93\u5165\u5e8f\u5217\uff0c\u800c\u4e0d\u4f1a\u88ab\u6b65\u9aa4\u6807\u8bb0\u7b26\u5e72\u6270\uff0c\u4ece\u800c\u66f4\u597d\u5730\u8bc6\u522b\u548c\u5206\u7c7b\u6b63\u786e\u6027\u3002<\/section>\n<section><\/section>\n<section><strong>LLM \u7684\u7b56\u7565\u5b66\u4e60\u3002<\/strong>\u6211\u4eec\u5c06\u6570\u5b66\u95ee\u9898\u8f6c\u6362\u4e3a\u4e00\u4e2a\u8bed\u8a00\u589e\u5f3a\u7684\u51b3\u7b56\u8fc7\u7a0b\uff0c\u7528\u6765\u9010\u6b65\u89e3\u51b3\u95ee\u9898\u3002\u8fd9\u4e2a\u8fc7\u7a0b\u53eb\u505a\u9a6c\u5c14\u53ef\u592b\u51b3\u7b56\u8fc7\u7a0b (MDP)\uff0c\u5b83\u7531\u72b6\u6001\u3001\u52a8\u4f5c\u548c\u5956\u52b1\u7ec4\u6210\u3002\u5728\u8fd9\u4e2a\u6846\u67b6\u4e2d\uff0c\u6bcf\u4e00\u4e2a\u6570\u5b66\u95ee\u9898\u5c31\u662f\u521d\u59cb\u72b6\u6001\uff0c\u6a21\u578b\u751f\u6210\u63a8\u7406\u6b65\u9aa4\u4f5c\u4e3a\u52a8\u4f5c\uff0c\u7136\u540e\u6839\u636e\u5f53\u524d\u72b6\u6001\u548c\u52a8\u4f5c\u6765\u51b3\u5b9a\u4e0b\u4e00\u4e2a\u72b6\u6001\u3002<\/section>\n<section><\/section>\n<section>\u6a21\u578b\u6bcf\u5b8c\u6210\u4e00\u4e2a\u6b65\u9aa4\uff0c\u5c31\u4f1a\u5f97\u5230\u4e00\u4e2a\u5956\u52b1\u6216\u53cd\u9988\uff0c\u7528\u6765\u8bc4\u4f30\u8be5\u6b65\u9aa4\u662f\u5426\u6b63\u786e\u3002\u8fd9\u4e2a\u5956\u52b1\u5e2e\u52a9\u6a21\u578b\u5224\u65ad\u662f\u5426\u671d\u7740\u6b63\u786e\u65b9\u5411\u524d\u8fdb\u3002\u6574\u4e2a\u8fc7\u7a0b\u91cd\u590d\u8fdb\u884c\uff0c\u6a21\u578b\u4f1a\u4e0d\u65ad\u8c03\u6574\u5176\u63a8\u7406\u8def\u5f84\uff0c\u76ee\u6807\u662f\u83b7\u5f97\u5c3d\u53ef\u80fd\u591a\u7684\u6b63\u9762\u53cd\u9988\u6216\u5956\u52b1\u3002<\/section>\n<section><\/section>\n<section>\u6211\u4eec\u5c06\u8fd9\u79cd MDP \u5b9e\u73b0\u4e3a\u4e00\u4e2a\u5f3a\u5316\u5b66\u4e60\u73af\u5883\uff0c\u7c7b\u4f3c OpenAI \u7684 Gym \u73af\u5883\u3002\u5728\u8fd9\u91cc\uff0c\u6bcf\u4e2a\u6570\u5b66\u95ee\u9898\u90fd\u88ab\u770b\u4f5c\u4e00\u4e2a\u4efb\u52a1\uff0c\u6a21\u578b\u901a\u8fc7\u4e00\u7cfb\u5217\u8fde\u7eed\u7684\u63a8\u7406\u6b65\u9aa4\u6765\u89e3\u51b3\u8fd9\u4e9b\u95ee\u9898\u3002\u6b63\u786e\u7684\u6b65\u9aa4\u83b7\u5f97\u5956\u52b1\uff0c\u9519\u8bef\u7684\u6b65\u9aa4\u5219\u53d7\u5230\u60e9\u7f5a\u3002\u901a\u8fc7\u8fd9\u79cd\u65b9\u5f0f\uff0c\u6a21\u578b\u53ef\u4ee5\u5728\u4e0d\u65ad\u8bd5\u9519\u4e2d\u4f18\u5316\u5176\u7b56\u7565\uff0c\u4ece\u800c\u9010\u6e10\u63d0\u9ad8\u5176\u89e3\u51b3\u6570\u5b66\u95ee\u9898\u7684\u80fd\u529b\u3002<\/section>\n<section><\/section>\n<section>\u5728\u7ebf\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\u3002\u5728\u4f7f\u7528\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\u5927\u578b\u8bed\u8a00\u6a21\u578b (LLM) \u65f6\uff0c\u901a\u5e38\u4f7f\u7528\u8fd1\u7aef\u7b56\u7565\u4f18\u5316 (PPO) \u6765\u4f7f\u751f\u6210\u7684\u8bed\u8a00\u8f93\u51fa\u4e0e\u9884\u671f\u7684\u52a8\u4f5c\u5bf9\u9f50\u3002PPO \u53ef\u4ee5\u5e2e\u52a9\u6a21\u578b\u751f\u6210\u65e2\u7b26\u5408\u8bed\u5883\u53c8\u8fbe\u5230\u76ee\u6807\u7684\u54cd\u5e94\uff0c\u586b\u8865\u4e86\u8bed\u8a00\u7406\u89e3\u548c\u64cd\u4f5c\u8f93\u51fa\u4e4b\u95f4\u7684\u7a7a\u9699\u3002\u6211\u4eec\u63d0\u4f9b\u4e86\u4f20\u7edf\u7684 PPO \u548c\u4e00\u79cd\u66f4\u9ad8\u6548\u7684\u53d8\u4f53\uff0c\u5373\u7fa4\u4f53\u76f8\u5bf9\u7b56\u7565\u4f18\u5316 (GRPO)\u3002\u8fd9\u4e24\u8005\u4e3b\u8981\u5728\u4f18\u52bf\u503c\u7684\u8ba1\u7b97\u65b9\u6cd5\u4e0a\u4e0d\u540c\uff1aPPO \u4f7f\u7528\u4e00\u4e2a\u7f51\u7edc\u6765\u4f30\u7b97\u72b6\u6001\u503c\uff0c\u5e76\u901a\u8fc7\u5e7f\u4e49\u4f18\u52bf\u4f30\u7b97 (GAE) \u6280\u672f\u6765\u8ba1\u7b97\u4f18\u52bf\u503c\uff1b\u800c GRPO \u5219\u7b80\u5316\u4e86\u8fd9\u4e2a\u8fc7\u7a0b\uff0c\u76f4\u63a5\u4f7f\u7528\u6807\u51c6\u5316\u7684\u5956\u52b1\u4fe1\u53f7\u6765\u4f30\u7b97\u52a8\u4f5c\u7684\u4f18\u52bf\uff0c\u4ece\u800c\u51cf\u5c11\u4e86\u8bad\u7ec3\u8d44\u6e90\u7684\u6d88\u8017\uff0c\u540c\u65f6\u66f4\u52a0\u6ce8\u91cd\u5956\u52b1\u6a21\u578b\u7684\u7a33\u5b9a\u6027\u3002<\/section>\n<section><\/section>\n<section><strong>\u89e3\u7801\uff1a\u63a8\u7406\u65f6\u7684\u5f15\u5bfc\u641c\u7d22\u548c\u89c4\u5212<\/strong><\/section>\n<section><\/section>\n<section>\u6211\u4eec\u4f7f\u7528 PRM \u6765\u8bc4\u4f30\u6bcf\u4e2a\u89e3\u51b3\u6b65\u9aa4\u7684\u51c6\u786e\u6027\u3002\u4e00\u65e6\u8bad\u7ec3\u51fa\u9ad8\u8d28\u91cf\u7684\u8fc7\u7a0b\u5956\u52b1\u6a21\u578b\uff0c\u6211\u4eec\u5c31\u53ef\u4ee5\u5c06\u5176\u4e0e\u8bed\u8a00\u6a21\u578b\u7ed3\u5408\u5230\u89e3\u7801\u8fc7\u7a0b\u4e2d\uff0c\u4ece\u800c\u5b9e\u73b0\u5f15\u5bfc\u641c\u7d22\u548c\u591a\u6b21\u751f\u6210\u7684\u8bc4\u5206\u6216\u6295\u7968\u3002<\/section>\n<section><\/section>\n<section>\u4e3a\u4e86\u5c06 PRM \u7528\u4f5c\u9a8c\u8bc1\u5668\uff0c\u6211\u4eec\u5b9a\u4e49\u4e86\u8bc4\u4f30 LLM \u751f\u6210\u7684\u89e3\u51b3\u65b9\u6848\u6b63\u786e\u6027\u7684\u65b9\u6cd5\uff0c\u5c06\u6bcf\u4e00\u6b65\u7684\u5f97\u5206\u8f6c\u6362\u4e3a\u6700\u7ec8\u5206\u6570\u3002\u4e3b\u8981\u6709\u4e24\u79cd\u65b9\u6cd5\uff1a<\/section>\n<section><\/section>\n<ul>\n<li>\n<section>PRM-Min\uff1a\u9009\u62e9\u6240\u6709\u6b65\u9aa4\u4e2d\u5f97\u5206\u6700\u4f4e\u7684\u4f5c\u4e3a\u6700\u7ec8\u5206\u6570\u3002<\/section>\n<\/li>\n<li>\n<section>PRM-Last\uff1a\u9009\u62e9\u6700\u540e\u4e00\u6b65\u7684\u5f97\u5206\u4f5c\u4e3a\u6700\u7ec8\u5206\u6570\u3002\u8fd9\u79cd\u65b9\u6cd5\u5df2\u7ecf\u88ab\u8bc1\u660e\u6548\u679c\u4e0e PRM-Min \u76f8\u5f53\u3002<\/section>\n<\/li>\n<\/ul>\n<section><\/section>\n<section>\u5f53\u901a\u8fc7\u6269\u5927\u63a8\u7406\u65f6\u8ba1\u7b97\u751f\u6210\u591a\u4e2a\u7b54\u6848\u540e\uff0c\u6211\u4eec\u9700\u8981\u57fa\u4e8e\u5206\u6570\u9009\u62e9\u6700\u4f73\u7b54\u6848\u3002\u6211\u4eec\u91c7\u7528\u4e86\u4e09\u79cd\u7b56\u7565\uff1a<\/section>\n<section><\/section>\n<section>1. \u591a\u6570\u6295\u7968\uff1a\u901a\u8fc7\u7edf\u8ba1\u51fa\u73b0\u6700\u591a\u7684\u7b54\u6848\u4f5c\u4e3a\u6700\u7ec8\u7b54\u6848\u3002<\/section>\n<section>2. RM-Max\uff1a\u6839\u636e\u7ed3\u679c\u5956\u52b1\u6a21\u578b\uff0c\u9009\u62e9\u6700\u7ec8\u5956\u52b1\u6700\u9ad8\u7684\u7b54\u6848\u3002<\/section>\n<section>3. RM-Vote\uff1a\u6839\u636e\u7ed3\u679c\u5956\u52b1\u6a21\u578b\uff0c\u9009\u62e9\u5956\u52b1\u603b\u548c\u6700\u9ad8\u7684\u7b54\u6848\u3002<\/section>\n<section><\/section>\n<section>\u901a\u8fc7\u7ed3\u5408\u8fd9\u4e9b\u7b56\u7565\uff0c\u53ef\u4ee5\u5f62\u6210\u591a\u79cd\u52a0\u6743\u65b9\u6cd5\uff0c\u4f8b\u5982 PRM-Last-Max\uff0c\u5373\u4f7f\u7528 PRM-Last \u548c RM-Max \u7ec4\u5408\u8fdb\u884c\u9009\u62e9\u3002\u6211\u4eec\u7684\u6846\u67b6\u5141\u8bb8\u6211\u4eec\u5728\u591a\u79cd\u641c\u7d22\u7b97\u6cd5\u4e2d\u8fdb\u884c\u9009\u62e9\uff0c\u4f8b\u5982 Beam Search\u3001Best-of-N, \u8499\u7279\u5361\u6d1b\u6811\u641c\u7d22\u7b49\u3002\u6bcf\u79cd\u7b97\u6cd5\u5728 PRM \u7684\u8d28\u91cf\u4e0a\u6709\u5176\u72ec\u7279\u7684\u4f18\u52bf\u3002\u590d\u6742\u7684\u641c\u7d22\u7b97\u6cd5\u5728\u5904\u7406\u66f4\u96be\u7684\u4efb\u52a1\u65f6\u53ef\u80fd\u8868\u73b0\u66f4\u597d\uff0c\u800c\u7b80\u5355\u7684\u65b9\u6cd5\u5982\u6700\u4f73 N \u5219\u5e38\u80fd\u5728\u96be\u5ea6\u8f83\u4f4e\u7684\u60c5\u51b5\u4e0b\u8868\u73b0\u826f\u597d\u3002<\/section>\n<section><\/section>\n<section><strong>\u89e3\u7801\u9636\u6bb5\u7684 Scaling Law<\/strong><\/section>\n<section><\/section>\n<section>\u6211\u4eec\u89c2\u5bdf\u5230\u4e86\u548c OpenAI o1 \u4ee5\u53ca Deepmind \u8bba\u6587\u300aScaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters\u300b\u8d8b\u52bf\u76f8\u8fd1\u7684 Test-time Scaling Law\uff0c\u53c2\u89c1\uff1a<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-d072169ac221316fa1f071b30ffb6e96.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-d072169ac221316fa1f071b30ffb6e96.png\"><\/a><\/p>\n<section><em><sup>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u56fe 3 \u63a8\u65ad\u9636\u6bb5\u65b0\u7684\u7f29\u653e\u7387\u5b9e\u9a8c\u6548\u679c\u56fe<\/sup><\/em><\/section>\n<section><\/section>\n<section>\u56fe 3 (a) \u6bd4\u8f83\u4e86\u8fd9\u4e9b\u641c\u7d22\u548c\u6295\u7968\u65b9\u6cd5\u5728\u63a8\u7406\u8fc7\u7a0b\u4e2d\u7684\u6027\u80fd\u3002y \u8f74\u8868\u793a MATH500 \u6570\u636e\u96c6\u4e0a\u7684\u6d4b\u8bd5\u51c6\u786e\u7387\uff0c\u800c x \u8f74\u663e\u793a\u751f\u6210\u9884\u7b97\uff08\u6bcf\u4e2a\u95ee\u9898\u7684\u5e73\u5747\u6807\u8bb0\u6570\uff09\uff0c\u53cd\u6620\u4e86\u6bcf\u4e2a\u95ee\u9898\u7684\u8ba1\u7b97\u6d88\u8017\u6216\u6807\u8bb0\u4f7f\u7528\u60c5\u51b5\u3002\u8be5\u56fe\u8868\u660e\uff0c\u968f\u7740\u751f\u6210\u9884\u7b97\u7684\u589e\u52a0\uff0c\u6700\u4f73 N \u9009\u62e9\u548c\u675f\u641c\u7d22\u65b9\u6cd5\u7684\u6027\u80fd\u663e\u8457\u4f18\u4e8e\u591a\u6570\u6295\u7968\uff0c\u4e0e\u4e4b\u524d\u7684\u53d1\u73b0\u8868\u73b0\u51fa\u76f8\u4f3c\u7684\u6a21\u5f0f\u3002\u5728\u4f4e\u63a8\u7406\u65f6\u8ba1\u7b97\u9884\u7b97\u4e0b\uff0c\u6700\u4f73 N \u9009\u62e9\u65b9\u6cd5\u8868\u73b0\u4f18\u4e8e\u675f\u641c\u7d22\uff0c\u800c\u675f\u641c\u7d22\u5728\u8f83\u9ad8\u9884\u7b97\u4e0b\u53ef\u4ee5\u8fbe\u5230\u76f8\u540c\u7684\u6027\u80fd\u3002\u53e6\u4e00\u65b9\u9762\uff0c\u56fe (b) \u663e\u793a\u6211\u4eec\u7684 PRM (Math-aps) \u80fd\u5728\u6240\u6709\u6d4b\u8bd5\u7684\u8ba1\u7b97\u9884\u7b97\u4e0b\u8fbe\u5230\u6700\u9ad8\u7684\u6d4b\u8bd5\u51c6\u786e\u7387\u3002\u8fd9\u786e\u5b9e\u9a8c\u8bc1\u4e86\u6211\u4eec\u7684 PRM \u8bad\u7ec3\u80fd\u591f\u6709\u6548\u5730\u5b66\u4e60\u8fc7\u7a0b\u76d1\u7763\u3002<\/section>\n<section><\/section>\n<section>\u8be6\u7ec6\u7684\u6587\u6863\u7ed3\u675f\u3002OpenR \u652f\u6301\u4f7f\u7528\u51e0\u884c\u4ee3\u7801\u5373\u53ef\u5b9e\u73b0 PRM \u7684\u8bad\u7ec3\u3001\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\uff0c\u4ee5\u53ca\u4e0d\u540c\u7684\u89e3\u7801\u65b9\u6cd5\uff0c\u4f7f\u7528\u6237\u80fd\u591f\u65b9\u4fbf\u5730\u8fdb\u884c\u5b9e\u9a8c\u548c\u6d4b\u8bd5\u3002\u6211\u4eec\u8fd8\u63d0\u4f9b\u4e86\u8be6\u7ec6\u7684\u4ee3\u7801\u6587\u6863\u4f9b\u5927\u5bb6\u53c2\u8003\uff0c\u53c2\u89c1: https:\/\/openreasoner.github.io\/ \u3002\u6211\u4eec\u6240\u652f\u6301\u7684\u7b97\u6cd5\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-85c5392197453e3c155a537174f9cc95.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-85c5392197453e3c155a537174f9cc95.png\"><\/a><\/p>\n<section><em><sup>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0\u56fe 4 \u5f00\u6e90\u4ee3\u7801\u7b97\u6cd5\u5b9e\u73b0\u6846\u56fe<\/sup><\/em><\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-e6cb139e610f64de3705a794a0c87a29.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-e6cb139e610f64de3705a794a0c87a29.png\"><\/a><\/p>\n<section><em><sup>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0\u56fe 5 OpenR \u6280\u672f\u6587\u6863\u56fe<\/sup><\/em><\/section>\n<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-10-14-5\" target=\"_blank\">\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846 [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","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":"","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":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","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":[27],"tags":[],"class_list":["post-36087","post","type-post","status-publish","format-standard","hentry","category-news"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03 - \u4e00\u8d77AI\u6280\u672f<\/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:\/\/17aitech.com\/?p=36087\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=36087\",\"url\":\"https:\/\/17aitech.com\/?p=36087\",\"name\":\"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03 - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=36087#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=36087#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png\",\"datePublished\":\"2025-01-05T09:00:10+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=36087#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=36087\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=36087#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=36087#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/17aitech.com\/#website\",\"url\":\"https:\/\/17aitech.com\/\",\"name\":\"\u4e00\u8d77AI\u6280\u672f\",\"description\":\"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca\",\"alternateName\":\"\u4e00\u8d77AI\u6280\u672f\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/17aitech.com\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\",\"name\":\"AI\u5c0f\u52a9\u624b\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/image\/\",\"url\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"contentUrl\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"caption\":\"AI\u5c0f\u52a9\u624b\"},\"description\":\"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e\",\"url\":\"https:\/\/17aitech.com\/?page_id=33738&user=3\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03 - \u4e00\u8d77AI\u6280\u672f","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:\/\/17aitech.com\/?p=36087","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=36087","url":"https:\/\/17aitech.com\/?p=36087","name":"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03 - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=36087#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=36087#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png","datePublished":"2025-01-05T09:00:10+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=36087#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=36087"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=36087#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/10\/frc-76518be6071cfd62a7552e246fcd86ba.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=36087#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u9996\u4e2ao1\u590d\u73b0\u5f00\u6e90RL\u6846\u67b6OpenR\u6765\u4e86\uff0cUCL\u3001\u4e0a\u4ea4\u7b49\u9ad8\u6821\u8054\u5408\u56e2\u961f\u53d1\u5e03"}]},{"@type":"WebSite","@id":"https:\/\/17aitech.com\/#website","url":"https:\/\/17aitech.com\/","name":"\u4e00\u8d77AI\u6280\u672f","description":"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca","alternateName":"\u4e00\u8d77AI\u6280\u672f","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/17aitech.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d","name":"AI\u5c0f\u52a9\u624b","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/#\/schema\/person\/image\/","url":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","caption":"AI\u5c0f\u52a9\u624b"},"description":"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e","url":"https:\/\/17aitech.com\/?page_id=33738&user=3"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/36087","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=36087"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/36087\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=36087"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=36087"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=36087"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}