{"id":7090,"date":"2025-07-18T06:21:00","date_gmt":"2025-07-17T21:21:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7090"},"modified":"2025-07-12T15:27:00","modified_gmt":"2025-07-12T06:27:00","slug":"robust-multimodal-large-language-models-against-modality-conflict","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7090","title":{"rendered":"Robust Multimodal Large Language Models Against Modality Conflict"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Robust Multimodal Large Language Models Against Modality Conflict\u00a0<\/strong>[94.1]<br>\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u5927\u8a00\u8a9e\u30e2\u30c7\u30eb(MLLM)\u306f\u3001\u73fe\u5b9f\u306e\u30b7\u30ca\u30ea\u30aa\u306b\u304a\u3044\u3066\u5e7b\u899a\u3092\u8d77\u3053\u3059\u50be\u5411\u304c\u3042\u308b\u3002 \u6211\u3005\u306f\u3001MLLM\u3092\u30b8\u30ec\u30f3\u30de\u306b\u914d\u7f6e\u3057\u3001\u5e7b\u899a\u306b\u76f4\u63a5\u5c0e\u304f\u7570\u306a\u308b\u30e2\u30c0\u30ea\u30c6\u30a3\u304b\u3089\u306e\u5165\u529b\u306b\u304a\u3051\u308b\u56fa\u6709\u306e\u77db\u76fe\u306b\u3064\u3044\u3066\u7814\u7a76\u3059\u308b\u3002 \u30e2\u30c0\u30ea\u30c6\u30a3\u885d\u7a81\u306b\u3088\u308b\u5e7b\u899a\u3092\u7de9\u548c\u3059\u308b3\u3064\u306e\u65b9\u6cd5\u304c\u63d0\u6848\u3055\u308c\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2507.07151v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2507.07151v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 09 Jul 2025 11:18:38 GMT)<\/li>\n\n\n\n<li>MLLM\u7279\u6709\u306e\u30cf\u30eb\u30b7\u30cd\u30fc\u30b7\u30e7\u30f3\uff08\u30e2\u30c0\u30ea\u30c6\u30a3\u9593\u306e\u4e0d\u6574\u5408\u306b\u95a2\u9023\u3059\u308b\u3082\u306e\uff09\u306b\u5bfe\u3059\u308b\u5bfe\u7b56\u306e\u6574\u7406\u300cMultimodal Modality Conflict (MMMC) \u300d\u3068\u3044\u3046\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3082\u4f5c\u6210\u3057\u691c\u8a3c\u3002\u691c\u8a3c\u306e\u4e2d\u3067\u306f\u30d7\u30ed\u30f3\u30d7\u30c8\u30a8\u30f3\u30b8\u30cb\u30a2\u30ea\u30f3\u30b0\u3001SFT\u3001\u5f37\u5316\u5b66\u7fd2\u3067\u306e\u30cf\u30eb\u30b7\u30cd\u30fc\u30b7\u30e7\u30f3\u8efd\u6e1b\u3092\u8a66\u3057\u300cOur results show that the reinforcement learning method achieves the best performance in mitigating the hallucination under modality conflict, while the supervised fine- tuning method shows promising and stable performance.\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/zmzhang2000\/MMMC\">GitHub &#8211; zmzhang2000\/MMMC: Official repository for Robust Multimodal Large Language Models Against Modality Conflict<\/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":[184,251],"class_list":["post-7090","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-hallucination","tag-mllm"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7090","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=7090"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7090\/revisions"}],"predecessor-version":[{"id":7091,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7090\/revisions\/7091"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7090"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7090"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7090"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}