{"id":6011,"date":"2025-01-09T04:23:00","date_gmt":"2025-01-08T19:23:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6011"},"modified":"2025-01-09T04:23:00","modified_gmt":"2025-01-08T19:23:00","slug":"os-genesis-automating-gui-agent-trajectory-construction-via-reverse-task-synthesis","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6011","title":{"rendered":"OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis\u00a0<\/strong>[55.4]<br>\u30b0\u30e9\u30d5\u30a3\u30ab\u30eb\u30e6\u30fc\u30b6\u30a4\u30f3\u30bf\u30d5\u30a7\u30fc\u30b9(GUI)\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u306e\u305f\u3081\u306e\u65b0\u3057\u3044\u30c7\u30fc\u30bf\u5408\u6210\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u3042\u308bOS-Genesis\u3092\u63d0\u6848\u3059\u308b\u3002 \u4e8b\u524d\u306b\u5b9a\u7fa9\u3055\u308c\u305f\u30bf\u30b9\u30af\u306b\u983c\u308b\u4ee3\u308f\u308a\u306b\u3001OS-Genesis\u306f\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u304c\u307e\u305a\u74b0\u5883\u3092\u8a8d\u8b58\u3057\u3001\u30b9\u30c6\u30c3\u30d7\u30ef\u30a4\u30c9\u306a\u30a4\u30f3\u30bf\u30e9\u30af\u30b7\u30e7\u30f3\u3092\u5b9f\u884c\u3059\u308b\u3053\u3068\u3092\u53ef\u80fd\u306b\u3059\u308b\u3002 \u6b21\u306b\u3001\u751f\u6210\u3055\u308c\u305f\u8ecc\u9053\u306e\u54c1\u8cea\u3092\u4fdd\u8a3c\u3059\u308b\u305f\u3081\u306b\u8ecc\u9053\u5831\u916c\u30e2\u30c7\u30eb\u3092\u7528\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2412.19723v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2412.19723v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Fri, 27 Dec 2024 16:21:58 GMT)<\/li>\n\n\n\n<li>\u6025\u901f\u306b\u7814\u7a76\u304c\u9032\u3080GUI\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u958b\u767a\u306e\u305f\u3081\u306e\u5408\u6210\u30c7\u30fc\u30bf\u69cb\u7bc9\u624b\u6cd5\u306e\u63d0\u6848\u3001\u300cOS-Genesis begins by exploring the functionality of GUI environments through traversing interactive UI elements with actions (e g , CLICK). This forms the basis for reverse task synthesis, where observed states and actions are retroactively transformed into low-level instructions. These low-level instructions are then derived into high-level instructions, which can seed the collection of GUI trajectories.\u300d\u3068\u57fa\u790e\u30c7\u30fc\u30bf\u3092\u69cb\u7bc9\u3001Trajectory Reward Model\u3067\u54c1\u8cea\u3092\u4fdd\u8a3c\u3002\u300cBuilt upon GPT-4o, TRM aims to perform a graded evaluation with a reward score R \u2208 [1, 5] to assist in sampling for training.\u300d\u3068\u306e\u3053\u3068\u30fb\u30fb\u30fb\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/qiushisun.github.io\/OS-Genesis-Home\/\">OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis<\/a><\/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":[42,180,181,390],"class_list":["post-6011","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-autonomous-agent","tag-gui","tag-gui-agent","tag-synthetic-data"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6011","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=6011"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6011\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6011"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6011"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6011"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}