{"id":7014,"date":"2025-07-02T05:20:00","date_gmt":"2025-07-01T20:20:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7014"},"modified":"2025-06-28T22:24:01","modified_gmt":"2025-06-28T13:24:01","slug":"robust-reward-modeling-via-causal-rubrics","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7014","title":{"rendered":"Robust Reward Modeling via Causal Rubrics\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Robust Reward Modeling via Causal Rubrics\u00a0<\/strong>[46.4]<br>\u30ea\u30ef\u30fc\u30c9\u30e2\u30c7\u30eb(RM)\u306f\u3001\u4eba\u9593\u306e\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u306b\u3088\u3063\u3066LLM(Large Language Models)\u3092\u6574\u5217\u3055\u305b\u308b\u306e\u306b\u57fa\u672c\u7684\u3060\u304c\u3001\u5831\u916c\u306e\u30cf\u30c3\u30ad\u30f3\u30b0\u306b\u60a9\u307e\u3055\u308c\u308b\u3053\u3068\u304c\u591a\u3044\u3002 Crome\u306f\u3001\u5831\u916c\u306e\u30cf\u30c3\u30ad\u30f3\u30b0\u3092\u8efd\u6e1b\u3059\u308b\u305f\u3081\u306b\u8a2d\u8a08\u3055\u308c\u305f\u660e\u78ba\u306a\u56e0\u679c\u30e2\u30c7\u30eb\u306b\u57fa\u3065\u304f\u3001\u65b0\u3057\u3044\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308b\u3002 RewardBench\u306e\u6a19\u6e96\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u3092\u5927\u5e45\u306b\u4e0a\u56de\u308a\u3001\u5e73\u5747\u7cbe\u5ea6\u3092\u6700\u59275.4%\u5411\u4e0a\u3055\u305b\u3001\u7279\u5b9a\u306e\u30ab\u30c6\u30b4\u30ea\u30fc\u3067\u6700\u592713.2%\u30687.2%\u306e\u30b2\u30a4\u30f3\u3092\u9054\u6210\u3057\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2506.16507v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2506.16507v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 19 Jun 2025 17:59:47 GMT)<\/li>\n\n\n\n<li>reward\u30cf\u30c3\u30ad\u30f3\u30b0\u3078\u5bfe\u5fdc\u53ef\u80fd\u306a\u56e0\u679c\u6027\u3092\u5229\u7528\u3057\u305f\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3001Crome (Causally Robust Reward Modeling)\u306e\u63d0\u6848<\/li>\n\n\n\n<li>Google Deepmind\u306b\u3088\u308b\u6210\u679c\u3060\u304cChrome\u3068\u7d1b\u3089\u308f\u3057\u3044\u3088\u3046\u306a\u30fb\u30fb\u30fb<\/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":[576],"class_list":["post-7014","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-576"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7014","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=7014"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7014\/revisions"}],"predecessor-version":[{"id":7015,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7014\/revisions\/7015"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7014"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7014"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7014"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}