{"id":6028,"date":"2025-01-24T04:11:00","date_gmt":"2025-01-23T19:11:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6028"},"modified":"2025-01-24T04:11:00","modified_gmt":"2025-01-23T19:11:00","slug":"benchmarking-large-and-small-mllms","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6028","title":{"rendered":"Benchmarking Large and Small MLLMs\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Benchmarking Large and Small MLLMs&nbsp;<\/strong>[71.8]<br>\u5927\u898f\u6a21\u306a\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u8a00\u8a9e\u30e2\u30c7\u30eb(MLLM)\u306f\u3001\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u30b3\u30f3\u30c6\u30f3\u30c4\u306e\u7406\u89e3\u3068\u751f\u6210\u306b\u304a\u3044\u3066\u9855\u8457\u306a\u9032\u6b69\u3092\u9042\u3052\u3066\u3044\u308b\u3002 \u3057\u304b\u3057\u3001\u305d\u306e\u30c7\u30d7\u30ed\u30a4\u30e1\u30f3\u30c8\u306f\u3001\u9045\u3044\u63a8\u8ad6\u3001\u9ad8\u3044\u8a08\u7b97\u30b3\u30b9\u30c8\u3001\u30c7\u30d0\u30a4\u30b9\u4e0a\u306e\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u306b\u5bfe\u3059\u308b\u975e\u73fe\u5b9f\u6027\u306a\u3069\u3001\u91cd\u5927\u306a\u8ab2\u984c\u306b\u76f4\u9762\u3057\u3066\u3044\u308b\u3002 LLava\u30b7\u30ea\u30fc\u30ba\u30e2\u30c7\u30eb\u3068Phi-3-Vision\u306b\u3088\u3063\u3066\u5b9f\u8a3c\u3055\u308c\u305f\u5c0f\u3055\u306aMLLM\u306f\u3001\u3088\u308a\u9ad8\u901f\u306a\u63a8\u8ad6\u3001\u30c7\u30d7\u30ed\u30a4\u30e1\u30f3\u30c8\u30b3\u30b9\u30c8\u306e\u524a\u6e1b\u3001\u30c9\u30e1\u30a4\u30f3\u56fa\u6709\u306e\u30b7\u30ca\u30ea\u30aa\u3092\u6271\u3046\u80fd\u529b\u3092\u5099\u3048\u305f\u6709\u671b\u306a\u4ee3\u66ff\u624b\u6bb5\u3092\u63d0\u4f9b\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2501.04150v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2501.04150v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Sat, 04 Jan 2025 07:44:49 GMT)<\/li>\n\n\n\n<li>MLLM\u306e\u5305\u62ec\u7684\u8a55\u4fa1\u3002<\/li>\n\n\n\n<li>\u300cGPT-4o establishes a new standard for multimodal understanding and reasoning across diverse input types, setting a benchmark in versatility and cognitive capacity.\u300d\u306e\u307b\u304b\u3001\u300cAlthough LLaVA-NeXT and Phi-3-Vision excel in specialized recognition tasks, they exhibit limitations in advanced reasoning and temporal sequence processing.\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li>MS\u306e\u8abf\u67fb\u3067\u3082\u3042\u308a\u3001Phi4\u3067\u306e\u30a2\u30c3\u30d7\u30c7\u30fc\u30c8\u306b\u3082\u671f\u5f85\u3002<a href=\"https:\/\/huggingface.co\/microsoft\/phi-4\">microsoft\/phi-4 \u00b7 Hugging Face<\/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":[251,365,368,517,647],"class_list":["post-6028","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-mllm","tag-slm","tag-small-model","tag-517","tag-647"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6028","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=6028"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6028\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6028"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6028"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6028"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}