{"id":4796,"date":"2024-05-03T03:40:00","date_gmt":"2024-05-02T18:40:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4796"},"modified":"2024-05-03T03:40:00","modified_gmt":"2024-05-02T18:40:00","slug":"weak-to-strong-extrapolation-expedites-alignment","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4796","title":{"rendered":"Weak-to-Strong Extrapolation Expedites Alignment"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Weak-to-Strong Extrapolation Expedites Alignment\u00a0<\/strong>[135.1]<br>\u4eba\u9593\u306e\u55dc\u597d\u3068LLM\u306e\u6574\u5408\u6027\u3092\u9ad8\u3081\u308b\u305f\u3081\u306e\u7c21\u5358\u306aExPO\u6cd5\u3092\u63d0\u6848\u3059\u308b\u3002 AlpacaEval 2.0\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u306f\u3001ExPO\u304c\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u305f\u30e2\u30c7\u30eb\u306b\u3001\u3088\u308a\u597d\u307f\u306e\u5c11\u306a\u3044\u30c7\u30fc\u30bf\u3067\u5230\u9054\u3057\u3001\u5b8c\u5168\u306b\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u305f\u30c7\u30fc\u30bf\u3092\u8d85\u3048\u3066\u3044\u308b\u3053\u3068\u304c\u793a\u3055\u308c\u3066\u3044\u307e\u3059\u3002 \u672c\u7814\u7a76\u306f,LLM\u306e\u80fd\u529b\u3092\u5229\u7528\u3057\u305f\u30e2\u30c7\u30eb\u5916\u633f\u306e\u6709\u52b9\u6027\u3092\u5b9f\u8a3c\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2404.16792v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2404.16792v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 25 Apr 2024 17:39:50 GMT)<\/li>\n\n\n\n<li>\u300cBy extrapolating from the weights of an SFT model Mw and a further trained one M, EXPO enables directly obtaining a better-aligned model without any additional training.\u300d\u3068\u3044\u3046\u624b\u6cd5\u306e\u63d0\u6848\u3002\u3068\u3066\u3082\u30b7\u30f3\u30d7\u30eb\u306b\u5916\u88c5\u3057\u3066\u3044\u308b\u3088\u3046\u306b\u898b\u3048\u3001\u306a\u3093\u3067\u3053\u3093\u306a\u3093\u3067\u3046\u3054\u304f\u3093\u3084\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/chujiezheng\/LLM-Extrapolation\">GitHub &#8211; chujiezheng\/LLM-Extrapolation: Official repository for paper &#8220;Weak-to-Strong Extrapolation Expedites Alignment&#8221;<\/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":[223,443,551],"class_list":["post-4796","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-weak-to-strong","tag-551"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4796","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=4796"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4796\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4796"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4796"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4796"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}