{"id":5814,"date":"2024-11-25T03:50:00","date_gmt":"2024-11-24T18:50:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5814"},"modified":"2024-11-25T03:50:00","modified_gmt":"2024-11-24T18:50:00","slug":"is-your-llm-secretly-a-world-model-of-the-internet-model-based-planning-for-web-agents","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5814","title":{"rendered":"Is Your LLM Secretly a World Model of the Internet? Model-Based Planning for Web Agents"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Is Your LLM Secretly a World Model of the Internet? Model-Based Planning for Web Agents\u00a0<\/strong>[23.2]<br>\u30e2\u30c7\u30eb\u30d9\u30fc\u30b9\u30d7\u30e9\u30f3\u30cb\u30f3\u30b0\u3067\u8a00\u8a9e\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u3092\u5897\u5f37\u3059\u308b\u65b0\u3057\u3044\u30d1\u30e9\u30c0\u30a4\u30e0\u3092\u5c0e\u5165\u3059\u308b\u3002 \u6211\u3005\u306e\u65b9\u6cd5\u3067\u3042\u308bWebDreamer\u306f\u3001LLM\u304c\u672c\u8cea\u7684\u306b\u30a6\u30a7\u30d6\u30b5\u30a4\u30c8\u306e\u69cb\u9020\u3084\u6a5f\u80fd\u306b\u95a2\u3059\u308b\u5305\u62ec\u7684\u77e5\u8b58\u3092\u30a8\u30f3\u30b3\u30fc\u30c9\u3057\u3066\u3044\u308b\u3068\u3044\u3046\u30ad\u30fc\u30a4\u30f3\u30b5\u30a4\u30c8\u3092\u69cb\u7bc9\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2411.06559v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2411.06559v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Sun, 10 Nov 2024 18:50:51 GMT)<\/li>\n\n\n\n<li>\u300cWEBDREAMER uses LLMs to simulate outcomes for each candidate action (e g , \u201cwhat would happen if I click this button?\u201d) using natural language descriptions, and then evaluates these imagined outcomes to determine the optimal action at each step.\u300d\u3068\u3044\u3046\u30b7\u30f3\u30d7\u30eb\u306a\u624b\u6cd5\u3067\u300cour model-based planning approach, WEBDREAMER, shows substantial improvement over reactive baselines and offers greater flexibility than tree search, which is often impossible in real-world websites.\u300d\u3068\u3044\u3046\u8208\u5473\u6df1\u3044\u7d50\u679c\u3002\u6311\u6226\u7684\u306a\u30bf\u30a4\u30c8\u30eb\u3092\u3064\u3051\u305f\u304f\u306a\u308b\u6c17\u6301\u3061\u3082\u308f\u304b\u308b\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/OSU-NLP-Group\/WebDreamer\/blob\/main\/README.md\">WebDreamer\/README.md at main \u00b7 OSU-NLP-Group\/WebDreamer \u00b7 GitHub<\/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":[223,449],"class_list":["post-5814","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-world-model"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5814","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=5814"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5814\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5814"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5814"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5814"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}