{"id":6383,"date":"2025-03-12T04:53:00","date_gmt":"2025-03-11T19:53:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6383"},"modified":"2025-03-12T04:53:00","modified_gmt":"2025-03-11T19:53:00","slug":"an-empirical-study-on-eliciting-and-improving-r1-like-reasoning-models","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6383","title":{"rendered":"An Empirical Study on Eliciting and Improving R1-like Reasoning Models\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>An Empirical Study on Eliciting and Improving R1-like Reasoning Models\u00a0<\/strong>[90.5]<br>RL\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u306f\u3001\u305d\u306e\u3088\u3046\u306a\u63a8\u8ad6\u30e2\u30c7\u30eb\u3092\u5b9f\u88c5\u3059\u308b\u305f\u3081\u306e\u4e2d\u5fc3\u7684\u306a\u30c6\u30af\u30cb\u30c3\u30af\u3068\u306a\u3063\u3066\u3044\u308b\u3002 \u6211\u3005\u306eRL\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30a2\u30d7\u30ed\u30fc\u30c1\u306fQwen2.5-32B\u30d9\u30fc\u30b9\u30e2\u30c7\u30eb\u3092\u7d99\u7d9a\u7684\u306b\u6539\u5584\u3059\u308b\u3053\u3068\u3092\u793a\u3057\u305f\u3002 \u307e\u305f\u3001\u30c4\u30fc\u30eb\u64cd\u4f5c\u306e\u5229\u7528\u306b\u3064\u3044\u3066\u3082\u691c\u8a0e\u3057\u3001\u5927\u304d\u306a\u63a8\u8ad6\u30e2\u30c7\u30eb\u306e\u63a8\u8ad6\u6027\u80fd\u3092\u5927\u5e45\u306b\u5411\u4e0a\u3055\u305b\u308b\u3053\u3068\u3092\u898b\u51fa\u3057\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2503.04548v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2503.04548v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 06 Mar 2025 15:34:27 GMT)<\/li>\n\n\n\n<li>\u69d8\u3005\u306a\u7814\u7a76\u6a5f\u95a2\u304c\u53d6\u308a\u7d44\u3080R1 like\uff08o1 like\uff09\u306a\u30e2\u30c7\u30eb\u958b\u767a\u306e\u30c6\u30af\u30cb\u30ab\u30eb\u30ec\u30dd\u30fc\u30c8\u3002\u300cBy effectively utilizing tool manipulation, STILL-3-TOOL-32B achieves an impressive accuracy of 86.67 (greedy search) on AIME 2024.  Remarkably, this ability can be activated with only a small number of high-quality training instances \u300d\u3068\u3044\u3046\u306e\u306f\u9762\u767d\u304f\u3001\u30c4\u30fc\u30eb\u306e\u5229\u7528\u306b\u3082\u62e1\u5f35\u304c\u9032\u307f\u3064\u3064\u3042\u308b\u3088\u3046\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/RUCAIBox\/Slow_Thinking_with_LLMs\">GitHub &#8211; RUCAIBox\/Slow_Thinking_with_LLMs: A series of technical report on Slow Thinking with LLM<\/a><\/li>\n<\/ul>\n\n\n\n<p><\/p>\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":[232,366],"class_list":["post-6383","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-lrm","tag-slow-thinking"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6383","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=6383"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6383\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6383"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6383"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6383"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}