{"id":3883,"date":"2023-10-02T06:01:00","date_gmt":"2023-10-01T21:01:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3883"},"modified":"2023-10-02T06:01:00","modified_gmt":"2023-10-01T21:01:00","slug":"llava-rlhf","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3883","title":{"rendered":"LLaVA-RLHF"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Aligning Large Multimodal Models with Factually Augmented RLHF\u00a0<\/strong>[176.5]<br>\u5927\u898f\u6a21\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u30e2\u30c7\u30eb(LMM)\u306f\u30e2\u30c0\u30ea\u30c6\u30a3\u306b\u307e\u305f\u304c\u3063\u3066\u69cb\u7bc9\u3055\u308c\u30012\u3064\u306e\u30e2\u30c0\u30ea\u30c6\u30a3\u9593\u306e\u30df\u30b9\u30a2\u30e9\u30a4\u30e1\u30f3\u30c8\u306f\u300challucination\u300d\u3092\u3082\u305f\u3089\u3059\u3002 \u30c6\u30ad\u30b9\u30c8\u9818\u57df\u304b\u3089\u8996\u899a\u8a00\u8a9e\u30a2\u30e9\u30a4\u30e1\u30f3\u30c8\u306e\u30bf\u30b9\u30af\u307e\u3067,RLHF(Reinforcement Learning from Human Feedback)\u3092\u9069\u5fdc\u3055\u305b\u308b\u3002 \u672c\u7a3f\u3067\u306f\u3001\u5831\u916c\u30e2\u30c7\u30eb\u306b\u4ed8\u52a0\u7684\u306a\u4e8b\u5b9f\u60c5\u5831\u3092\u8ffd\u52a0\u3059\u308bFactually Augmented RLHF\u3068\u3044\u3046\u65b0\u3057\u3044\u30a2\u30e9\u30a4\u30e1\u30f3\u30c8\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u63d0\u6848\u3059\u308b\u3002 \u63d0\u6848\u624b\u6cd5\u306f,\u30c6\u30ad\u30b9\u30c8\u306e\u307f\u306eGPT-4\u306e\u6027\u80fd\u30ec\u30d9\u30eb\u304c94%\u3067\u3042\u308bLLaVA-Bench\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u304a\u3044\u3066,\u9855\u8457\u306a\u6539\u5584\u3092\u5b9f\u73fe\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2309.14525v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2309.14525v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 25 Sep 2023 20:59:33 GMT)<\/li>\n\n\n\n<li>\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u30e2\u30c7\u30eb\u306b\u5bfe\u3059\u308bRLHF\u3068\u3057\u3066Factually Augmented RLHF (Fact-RLHF)\u00a0\u3092\u63d0\u6848\u3002\u30e2\u30c7\u30eb\u304c\u516c\u958b\u3055\u308c\u3066\u3044\u308b\u306e\u304c\u8208\u5473\u6df1\u304fGPT-4V\uff08<a href=\"https:\/\/openai.com\/research\/gpt-4v-system-card\">GPT-4V(ision) system card (openai.com)<\/a>\uff09\u3068\u6bd4\u8f03\u3057\u3066\u307f\u305f\u3068\u3053\u308d\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/llava-rlhf.github.io\/\">LLaVA-RLHF<\/a>\u3002\u30e2\u30c7\u30eb\u306f<a href=\"https:\/\/huggingface.co\/zhiqings\/LLaVA-RLHF-13b-v1.5-336\">zhiqings\/LLaVA-RLHF-13b-v1.5-336 \u00b7 Hugging Face<\/a>\uff08\u00a0Apache License 2.0\uff09\u306a\u3069\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":[22,223,340,524],"class_list":["post-3883","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-alignment","tag-llm","tag-rlhf","tag-524"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3883","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=3883"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3883\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3883"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3883"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3883"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}