{"id":7528,"date":"2025-10-09T04:46:00","date_gmt":"2025-10-08T19:46:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7528"},"modified":"2025-10-04T16:50:23","modified_gmt":"2025-10-04T07:50:23","slug":"infoagent-advancing-autonomous-information-seeking-agents","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7528","title":{"rendered":"InfoAgent: Advancing Autonomous Information-Seeking Agents"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>InfoAgent: Advancing Autonomous Information-Seeking Agents\u00a0<\/strong>[143.2]<br>\u672c\u7a3f\u3067\u306f,\u9769\u65b0\u7684\u306a\u30c7\u30fc\u30bf\u5408\u6210\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3068Web\u691c\u7d22\u30c4\u30fc\u30eb\u3092\u99c6\u4f7f\u3057\u305f\u30c7\u30a3\u30fc\u30d7\u30ea\u30b5\u30fc\u30c1\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8InfoAgent\u3092\u7d39\u4ecb\u3059\u308b\u3002 \u6211\u3005\u306e\u65b9\u6cd5\u3067\u306f\u3001InfoAgent\u306fBrowseComp\u306715.3%\u3001BrowseComp-ZH\u306729.2%\u3001Xbench-DS\u306740.4%\u306e\u7cbe\u5ea6\u3092\u9054\u6210\u3057\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2509.25189v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2509.25189v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 29 Sep 2025 17:59:57 GMT)<\/li>\n\n\n\n<li>Deep Research\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u306e\u69cb\u7bc9\u3002Qwen3 14B\u30d9\u30fc\u30b9\u3067\u5408\u6210\u30c7\u30fc\u30bf\u3092\u6d3b\u7528\u3001\u300cIn the first stage, we perform supervised finetuning (SFT) as a cold start, in order to instill long-horizon search behavior into the model.\u300d\u3001\u300cIn the second stage, we apply RL to refine its ability of reasoning-driven tool use.\u300d\u306e2\u6bb5\u968e\u3067\u306epost training\u3002<\/li>\n\n\n\n<li>\u5408\u6210\u30c7\u30fc\u30bf\u3001post training\u306e\u6709\u52b9\u6027\u3092\u793a\u3059\u7d50\u679c\u3067\u3001\u30d9\u30fc\u30b9\u30e2\u30c7\u30eb\u30b5\u30a4\u30ba\u3082\u304a\u624b\u9803\u611f\u304c\u3042\u308b\u3002\u3053\u306e\u3088\u3046\u306aSLM\u306e\u958b\u767a\u304c\u6d41\u884c\u3063\u3066\u3044\u304f\u53ef\u80fd\u6027\u3092\u611f\u3058\u308b\u7d50\u679c\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-7528","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\/7528","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=7528"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7528\/revisions"}],"predecessor-version":[{"id":7529,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7528\/revisions\/7529"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7528"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7528"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7528"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}