{"id":5618,"date":"2024-10-16T05:51:00","date_gmt":"2024-10-15T20:51:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5618"},"modified":"2024-10-16T05:51:00","modified_gmt":"2024-10-15T20:51:00","slug":"llms-are-in-context-reinforcement-learners","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5618","title":{"rendered":"LLMs Are In-Context Reinforcement Learners"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>LLMs Are In-Context Reinforcement Learners\u00a0<\/strong>[30.2]<br>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u306f\u3001\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u5185\u6559\u5e2b\u3042\u308a\u5b66\u7fd2(ICL)\u3092\u901a\u3058\u3066\u65b0\u3057\u3044\u30bf\u30b9\u30af\u3092\u5b66\u7fd2\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u3002 \u3053\u306e\u7814\u7a76\u306f\u3001\u3053\u306e\u80fd\u529b\u304c\u6587\u8108\u5185\u5f37\u5316\u5b66\u7fd2(ICRL)\u306b\u307e\u3067\u62e1\u5f35\u3055\u308c\u308b\u304b\u3069\u3046\u304b\u3092\u7814\u7a76\u3059\u308b\u3002 \u672c\u7a3f\u3067\u306f\u3001\u30c6\u30b9\u30c8\u6642\u9593\u8a08\u7b97\u306e\u5897\u52a0\u3068\u8a08\u7b97\u30d0\u30a6\u30f3\u30c9\u8fd1\u4f3c\u306b\u3088\u308a\u3001\u3053\u306e\u6b20\u9665\u306b\u5bfe\u51e6\u3059\u308b\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2410.05362v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2410.05362v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 07 Oct 2024 17:45:00 GMT)<\/li>\n\n\n\n<li>\u300cICRL is a natural combination of ICL and reinforcement learning (RL).Instead of constructing the LLM context from supervised input-output pairs, the LLM context is constructed using triplets consisting of input, model output prediction, and the corresponding rewards.\u300d\u3068\u3044\u3046\u30b9\u30bf\u30a4\u30eb\u306e\u30a4\u30f3\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u5f37\u5316\u5b66\u7fd2\u306e\u63d0\u6848\u3002\u30ca\u30a4\u30fc\u30d6\u306a\u5b9f\u88c5\u304c\u3046\u307e\u304f\u3044\u304b\u306a\u3044\u306e\u304c\u8208\u5473\u6df1\u3044\u3002\u300cIts poor performance is due to its incapacity to explore the output space.\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/lil-lab.github.io\/icrl\/\">LLMs Are In-Context Reinforcement Learners (lil-lab.github.io)<\/a><\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><\/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":[200,576],"class_list":["post-5618","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-in-context-learning","tag-576"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5618","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=5618"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5618\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5618"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5618"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5618"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}