{"id":7427,"date":"2025-09-18T05:16:00","date_gmt":"2025-09-17T20:16:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7427"},"modified":"2025-09-14T09:19:03","modified_gmt":"2025-09-14T00:19:03","slug":"sfr-deepresearch-towards-effective-reinforcement-learning-for-autonomously-reasoning-single-agents","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7427","title":{"rendered":"SFR-DeepResearch: Towards Effective Reinforcement Learning for Autonomously Reasoning Single Agents"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>SFR-DeepResearch: Towards Effective Reinforcement Learning for Autonomously Reasoning Single Agents\u00a0<\/strong>[93.3]<br>\u672c\u7a3f\u3067\u306f,\u30c7\u30a3\u30fc\u30d7\u30ea\u30b5\u30fc\u30c1\u306e\u305f\u3081\u306e\u30cd\u30a4\u30c6\u30a3\u30d6\u81ea\u5f8b\u5358\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30e2\u30c7\u30eb\u306e\u958b\u767a\u306b\u7126\u70b9\u3092\u3042\u3066\u308b\u3002 \u6211\u3005\u306e\u6700\u826f\u306e\u5909\u7a2e\u3067\u3042\u308bSFR-DR-20B\u306f\u3001Humanity\u306eLast Exam\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u306728.7%\u306b\u9054\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2509.06283v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2509.06283v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 08 Sep 2025 02:07:09 GMT)<\/li>\n\n\n\n<li>\u300cwe propose a compact synthetic-data reinforcement learning recipe that adapts reasoningoptimized LLMs into native Autonomous Single-Agent systems for Deep Research. Applied to open-source backbones, our best variant attains 28.7% on Humanity\u2019s Last Exam.\u300d\u3068\u5408\u6210\u30c7\u30fc\u30bf\u3092\u6d3b\u7528\u3057\u305fDeep Research\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u69cb\u7bc9\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306e\u63d0\u6848\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":[686,390],"class_list":["post-7427","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-deepresearch","tag-synthetic-data"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7427","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=7427"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7427\/revisions"}],"predecessor-version":[{"id":7428,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7427\/revisions\/7428"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7427"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7427"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7427"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}