{"id":5544,"date":"2024-10-07T06:45:00","date_gmt":"2024-10-06T21:45:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5544"},"modified":"2024-10-07T06:45:00","modified_gmt":"2024-10-06T21:45:00","slug":"%e6%97%a5%e6%9c%ac%e8%aa%9e%e7%89%88gemma-2-2b-liquid-foundation-models-lfms","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5544","title":{"rendered":"\u65e5\u672c\u8a9e\u7248Gemma 2 2B, Liquid Foundation Models (LFMs), Meta Movie Gen, CulturalBench"},"content":{"rendered":"\n<p>\u5148\u9031\u306e\u767a\u8868\u3067\u6c17\u306b\u306a\u3063\u305f\u306e\u306fGoogle\u306b\u3088\u308b\u65e5\u672c\u8a9e\u7248 Gemma 2\u306e\u516c\u958b\uff08<a href=\"https:\/\/developers-jp.googleblog.com\/2024\/10\/gemma-2-for-japan.html\">Google Developers Japan: \u65e5\u672c\u8a9e\u7248 Gemma 2 2B \u3092\u516c\u958b (googleblog.com)<\/a>\uff09\u3068Liquid AI\u306b\u3088\u308bLiquid Foundation Models (LFMs)\u306e\u767a\u8868\uff08<a href=\"https:\/\/www.liquid.ai\/liquid-foundation-models\">Liquid Foundation Models: Our First Series of Generative AI Models<\/a>\uff09\u3001Meta\u306b\u3088\u308b\u52d5\u753b\u751f\u6210AI\u3001Meta Movie Gen\u306e\u767a\u8868\uff08<a href=\"https:\/\/ai.meta.com\/research\/movie-gen\/\">Meta Movie Gen<\/a>\uff09\u3060\u3063\u305f\u3002<\/p>\n\n\n\n<p>\uff11\u3064\u76ee\u306f\u8a00\u8a9e\u7279\u5316\u30e2\u30c7\u30eb\u306e\u53ef\u80fd\u6027\u3092\u611f\u3058\u308b\u5c0f\u898f\u6a21\u30fb\u9ad8\u6027\u80fd\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002\u300c<a href=\"https:\/\/www.isct.ac.jp\/ja\">\u6771\u4eac\u79d1\u5b66\u5927\u5b66<\/a> \u60c5\u5831\u7406\u5de5\u5b66\u9662 \u60c5\u5831\u5de5\u5b66\u7cfb\u306e<a href=\"https:\/\/www.chokkan.org\/index.ja.html\">\u5ca1\u5d0e\u76f4\u89b3\u6559\u6388<\/a>\u3089\u306e\u7814\u7a76\u30c1\u30fc\u30e0\u3068\u5354\u529b\u3057\u3001\u65e5\u672c\u306b\u304a\u3051\u308b\u30aa\u30fc\u30d7\u30f3\u30e2\u30c7\u30eb\u306e\u958b\u767a\u652f\u63f4\u3001\u304a\u3088\u3073\u3001\u65b0\u3057\u3044\u6280\u8853\u306e\u958b\u62d3\u3078\u306e\u53d6\u308a\u7d44\u307f\u3082\u9032\u3081\u307e\u3059\u3002\u300d\u3068\u306e\u8a18\u8f09\u3082\u3042\u308a\u3001\u65e5\u672c\u8a9e\u3068\u3044\u3046\u8a00\u8a9e\u3060\u3051\u3067\u306a\u304f\u6587\u5316\u7406\u89e3\u306e\u3088\u3046\u306a\u90e8\u5206\u306b\u3082\u6ce8\u76ee\u3002\u5148\u9031\u3067\u3066\u3044\u305fCultualBench\u306e\u3088\u3046\u306a\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u69cb\u7bc9\u306e\u52d5\u304d\u3082\u3055\u304b\u3093\u3002<\/p>\n\n\n\n<p>\uff12\u3064\u76ee\u306fGPT\u7cfb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3067\u306f\u306a\u3044LLM\u3068\u306e\u3053\u3068\u3002\u8ad6\u6587\u3084\u30c6\u30af\u30cb\u30ab\u30eb\u30ec\u30dd\u30fc\u30c8\u304c\u51fa\u3066\u3044\u306a\u3044\u306e\u3067\u4f55\u3068\u3082\u8a00\u3048\u306a\u3044\u90e8\u5206\u304c\u3042\u308b\u304c\u3001\u72b6\u614b\u7a7a\u9593\u30e2\u30c7\u30eb\u3067\u306f\u306a\u304fAttention\u3092\u52b9\u7387\u5316\u3059\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u306e\u3088\u3046\u306b\u898b\u3048\u308b\u3002\u9577\u6587\u306b\u304a\u3051\u308b\u51e6\u7406\u304c\u5927\u5e45\u306b\u52b9\u7387\u5316\u3055\u308c\u3066\u3044\u308b\u3068\u306e\u3053\u3068\u3067\u671f\u5f85\u5927\u3002<\/p>\n\n\n\n<p>\u6700\u5f8c\u306fMeta\u306b\u3088\u308b\u30c6\u30ad\u30b9\u30c8\u304b\u3089\u306e\u52d5\u753b\u751f\u6210AI\u3067\u5358\u7d14\u306a\u751f\u6210\u3060\u3051\u3067\u306a\u304f\u3001\u7de8\u96c6\u3082\u53ef\u80fd\u3001\u5143\u306e\u9759\u6b62\u753b\u3082\u6307\u5b9a\u53ef\u80fd\u3002\u300cOn text-to-video generation, we outperform prior state-of-the-art, including commercial systems such as Runway Gen3 (RunwayML, 2024), LumaLabs (LumaLabs, 2024), OpenAI Sora (OpenAI, 2024) on overall video quality\u300d\u3068\u4ed6\u30e2\u30c7\u30eb\u3088\u308a\u3082\u826f\u3044\u6027\u80fd\u3067\u3042\u308b\u3068\u306e\u3053\u3068\u3002<br>\uff0810\/19\u8ffd\u8a18\uff09 arXiv\u306b\u8ad6\u6587\u304c\u51fa\u3066\u3044\u305f\u306e\u3067\u8ffd\u52a0\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>CulturalBench: a Robust, Diverse and Challenging Benchmark on Measuring the (Lack of) Cultural Knowledge of LLMs&nbsp;<\/strong>[75.8]<br>\u6587\u5316\u30d9\u30f3\u30c1\u306b\u3064\u3044\u3066\u7d39\u4ecb\u3059\u308b: \u6587\u5316\u7684\u77e5\u8b58\u3092\u8a55\u4fa1\u3059\u308b\u305f\u3081\u306e1,227\u306e\u4eba\u6587\u7684\u30fb\u4eba\u6587\u7684\u306a\u8cea\u554f\u3067\u3042\u308b\u3002 \u540c\u3058\u8cea\u554f\u3092\u5171\u6709\u3057\u306a\u304c\u3089\u7570\u306a\u308b\u8cea\u554f\u3092\u3059\u308bCulturalBench-Easy\u3068CulturalBench-Hard\u306e2\u3064\u306e\u8a2d\u5b9a\u3067\u30e2\u30c7\u30eb\u3092\u8a55\u4fa1\u3059\u308b\u3002 \u4eba\u9593\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9(92.6%\u306e\u7cbe\u5ea6)\u3068\u6bd4\u8f03\u3057\u3066\u3001\u30ab\u30eb\u30c1\u30e3\u30fc\u30d9\u30f3\u30c1\u30fb\u30cf\u30fc\u30c9\u306f\u3001\u6700\u3082\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u306e\u826f\u3044\u30e2\u30c7\u30eb(GPT-4o)\u304c61.5%\u3001\u6700\u60aa\u306e\u30e2\u30c7\u30eb(Llama3-8b)\u304c21.4%\u3067\u3042\u308b\u30d5\u30ed\u30f3\u30c6\u30a3\u30a2\u306eLLM\u306b\u3068\u3063\u3066\u3088\u308a\u96e3\u3057\u3044\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2410.02677v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2410.02677v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Thu, 03 Oct 2024 17:04:31 GMT)<\/li>\n\n\n\n<li>45\u304b\u56fd\u3092\u30ab\u30d0\u30fc\u3059\u308b\u6587\u5316\u7684\u30d9\u30f3\u30c1\u30de\u30fc\u30af<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/spaces\/kellycyy\/CulturalBench\">CulturalBench &#8211; a Hugging Face Space by kellycyy<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Movie Gen: A Cast of Media Foundation Models\u00a0<\/strong>[133.4]<br>\u9ad8\u54c1\u8cea\u306e1080pHD\u30d3\u30c7\u30aa\u3092\u751f\u6210\u3059\u308b\u57fa\u790e\u30e2\u30c7\u30eb\u306e\u30ad\u30e3\u30b9\u30c8\u3067\u3042\u308bMovie Gen\u306b\u3064\u3044\u3066\u7d39\u4ecb\u3059\u308b\u3002 \u30e6\u30fc\u30b6\u306e\u753b\u50cf\u306b\u57fa\u3065\u3044\u3066,\u9ad8\u7cbe\u5ea6\u306a\u547d\u4ee4\u30d9\u30fc\u30b9\u306e\u30d3\u30c7\u30aa\u7de8\u96c6\u3084\u30d1\u30fc\u30bd\u30ca\u30e9\u30a4\u30ba\u3055\u308c\u305f\u30d3\u30c7\u30aa\u306e\u751f\u6210\u306a\u3069\u306e\u8ffd\u52a0\u6a5f\u80fd\u3092\u793a\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2410.13720v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2410.13720v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 17 Oct 2024 16:22:46 GMT)<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u5148\u9031\u306e\u767a\u8868\u3067\u6c17\u306b\u306a\u3063\u305f\u306e\u306fGoogle\u306b\u3088\u308b\u65e5\u672c\u8a9e\u7248 Gemma 2\u306e\u516c\u958b\uff08Google Developers Japan: \u65e5\u672c\u8a9e\u7248 Gemma 2 2B \u3092\u516c\u958b (googleblog.com)\uff09\u3068Liquid AI &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=5544\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;\u65e5\u672c\u8a9e\u7248Gemma 2 2B, Liquid Foundation Models (LFMs), Meta Movie Gen, CulturalBench&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[223,517,554],"class_list":["post-5544","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-517","tag-554"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5544","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=5544"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5544\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5544"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5544"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5544"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}