{"id":6779,"date":"2025-05-20T04:13:00","date_gmt":"2025-05-19T19:13:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6779"},"modified":"2025-05-17T16:19:37","modified_gmt":"2025-05-17T07:19:37","slug":"j1-incentivizing-thinking-in-llm-as-a-judge-via-reinforcement-learning","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6779","title":{"rendered":"J1: Incentivizing Thinking in LLM-as-a-Judge via Reinforcement Learning\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>J1: Incentivizing Thinking in LLM-as-a-Judge via Reinforcement Learning\u00a0<\/strong>[69.1] <br>\u3053\u306e\u3088\u3046\u306a\u30e2\u30c7\u30eb\u3092\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u305f\u3081\u306e\u5f37\u5316\u5b66\u7fd2\u30a2\u30d7\u30ed\u30fc\u30c1\u3067\u3042\u308bJ1\u3092\u7d39\u4ecb\u3059\u308b\u3002 \u672c\u624b\u6cd5\u306f,\u5224\u65ad\u30d0\u30a4\u30a2\u30b9\u3092\u8efd\u6e1b\u3057,\u601d\u8003\u306b\u30a4\u30f3\u30bb\u30f3\u30c6\u30a3\u30d6\u3092\u4e0e\u3048\u308b\u691c\u8a3c\u53ef\u80fd\u306a\u5831\u916c\u3092\u7528\u3044\u3066,\u691c\u8a3c\u53ef\u80fd\u306a\u30d7\u30ed\u30f3\u30d7\u30c8\u3068\u691c\u8a3c\u4e0d\u53ef\u80fd\u306a\u30d7\u30ed\u30f3\u30d7\u30c8\u306e\u4e21\u65b9\u3092\u5224\u65ad\u30bf\u30b9\u30af\u306b\u5909\u63db\u3059\u308b\u3002 \u8a55\u4fa1\u57fa\u6e96\u3092\u6982\u8aac\u3057\u3001\u81ea\u5df1\u751f\u6210\u3057\u305f\u57fa\u6e96\u56de\u7b54\u3068\u6bd4\u8f03\u3057\u3001\u30e2\u30c7\u30eb\u5fdc\u7b54\u306e\u6b63\u3057\u3055\u3092\u518d\u8a55\u4fa1\u3059\u308b\u3053\u3068\u306b\u3088\u308a\u3001\u30e2\u30c7\u30eb\u304c\u3088\u308a\u826f\u3044\u5224\u65ad\u3092\u4e0b\u3059\u3053\u3068\u304c\u5224\u660e\u3057\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2505.10320v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2505.10320v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 15 May 2025 14:05:15 GMT)<\/li>\n\n\n\n<li>Thinking-LLM-as-a-Judge models\u3092\u69cb\u7bc9\u3059\u308b\u305f\u3081\u306e\u5f37\u5316\u5b66\u7fd2\u30ec\u30b7\u30d4\u306e\u63d0\u6848\u3002<\/li>\n\n\n\n<li>\u300cour approach outperforms all other existing 8B or 70B models when trained at those sizes, including models distilled from DeepSeek-R1. J1 also outperforms o1-mini, and even R1 on some benchmarks, despite training a smaller model.\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li><a href=\"https:\/\/devneko.jp\/wordpress\/?p=6614\">Assessing Judging Bias in Large Reasoning Models: An Empirical Study \u2013 arXiv\u6700\u65b0\u8ad6\u6587\u306e\u7d39\u4ecb<\/a>\u306a\u3069\u3001LLM as a judge\u306a\u30bf\u30b9\u30af\u3067\u306eLRM\u9069\u7528\u306b\u52b9\u679c\u304c\u3042\u308b\u3068\u3044\u3046\u6307\u6458\u306f\u3042\u3063\u305f\u306e\u3067\u305d\u308c\u3089\u3068\u6574\u5408\u7684\u306a\u7d50\u679c\u3067\u3042\u308b\u3088\u3046\u306b\u601d\u3046\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[224,232,576],"class_list":["post-6779","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm-as-a-judge","tag-lrm","tag-576"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6779","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6779"}],"version-history":[{"count":2,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6779\/revisions"}],"predecessor-version":[{"id":6781,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6779\/revisions\/6781"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6779"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6779"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6779"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}