{"id":5330,"date":"2024-08-26T05:57:00","date_gmt":"2024-08-25T20:57:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5330"},"modified":"2024-08-26T05:57:00","modified_gmt":"2024-08-25T20:57:00","slug":"hermes-3-jamba-1-5-mistral-nemo-minitron-8b-llama-3-1-minitron-4b","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5330","title":{"rendered":"HERMES 3, Jamba 1.5, Mistral-NeMo-Minitron-8B, Llama-3.1-Minitron-4B, BLIP-3"},"content":{"rendered":"\n<p>\u5148\u9031\u3082\u516c\u958b\u30e2\u30c7\u30eb\u306b\u95a2\u3059\u308b\u5831\u544a\u304c\u51fa\u3066\u3044\u305f\u3002HERMES 3\u306fHERMES\u30b7\u30ea\u30fc\u30ba\u306e\u6700\u65b0\u7248\u3067Llama3.1\u30d9\u30fc\u30b9(<a href=\"https:\/\/huggingface.co\/papers\/2408.11857\">Paper page &#8211; Hermes 3 Technical Report (huggingface.co)<\/a>)\u3001Jamba1.5\u306fTransformer-Mamba\u306e\u6df7\u5408\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3002&nbsp;Jamba-1.5-Large\u306f\u30a2\u30af\u30c6\u30a3\u30d6\u30d1\u30e9\u30e1\u30fc\u30bf\u304c94B\u3068\u3053\u306e\u624b\u306e\u30e2\u30c7\u30eb\u3067\u306f\u975e\u5e38\u306b\u5927\u898f\u6a21\u3002\u6700\u5f8c\u306eMinitron\u306f\u4ee5\u524d\u7d39\u4ecb\u3057\u305f\u30a2\u30d7\u30ed\u30fc\u30c1\u3067\u5c0f\u898f\u6a21\u5316\u3057\u305f\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Jamba-1.5: Hybrid Transformer-Mamba Models at Scale&nbsp;<\/strong>[30.9]<br>\u63d0\u6848\u3059\u308bJamba-1.5\u306f,Jamba\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u57fa\u3065\u304f\u547d\u4ee4\u8abf\u6574\u578b\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002 94B\u306e\u30a2\u30af\u30c6\u30a3\u30d6\u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u6301\u3064Jamba-1.5-Large\u306812B\u306e\u30a2\u30af\u30c6\u30a3\u30d6\u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u6301\u3064Jamba-1.5-Mini\u306e2\u3064\u306e\u30e2\u30c7\u30eb\u30b5\u30a4\u30ba\u3092\u30ea\u30ea\u30fc\u30b9\u3059\u308b\u3002 \u3069\u3061\u3089\u306e\u30e2\u30c7\u30eb\u3082\u69d8\u3005\u306a\u5bfe\u8a71\u578b\u304a\u3088\u3073\u547d\u4ee4\u8ffd\u5f93\u578b\u30ad\u30e3\u30d1\u30d3\u30ea\u30c6\u30a3\u306e\u305f\u3081\u306b\u5fae\u8abf\u6574\u3055\u308c\u3066\u304a\u308a\u3001\u6709\u52b9\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u9577\u306f256K\u3067\u3042\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2408.12570v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2408.12570v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Thu, 22 Aug 2024 17:38:59 GMT)<\/li>\n\n\n\n<li>Transformer\u3068Mamba\u306e\u6df7\u5408\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3067\u9ad8\u901f\u306a\u30e2\u30c7\u30ebJamba\u306e\u30a2\u30c3\u30d7\u30c7\u30fc\u30c8\u3002\u4ee5\u524d\u3088\u308a\u5927\u898f\u6a21\u5316\u3057\u3066\u3044\u308b\u3002Llama3.1\u3084Gemma\u3001Mistral\u306a\u3069\u306eTransformer\u30d9\u30fc\u30b9\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3068\u6bd4\u8f03\u3055\u308c\u3066\u3044\u308b\u304c\u3001\u30b9\u30b3\u30a2\u7279\u6027\u304c\u9055\u3046\u3088\u3046\u306b\u898b\u3048\u308b\u306e\u304c\u8208\u5473\u6df1\u3044\u3002\uff08\u30c7\u30fc\u30bf\u306e\u5dee\u304b\u3082\u3057\u308c\u306a\u3044\u304c\u3002\u3002\uff09<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/collections\/ai21labs\/jamba-15-66c44befa474a917fcf55251\">Jamba-1.5 &#8211; a ai21labs Collection (huggingface.co)<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>LLM Pruning and Distillation in Practice: The Minitron Approach&nbsp;<\/strong>[61.6]<br>Llama 3.1 8B \u304a\u3088\u3073 Mistral NeMo 12B \u30e2\u30c7\u30eb\u3092 4B \u304a\u3088\u3073 8B \u30d1\u30e9\u30e1\u30fc\u30bf\u306b\u5727\u7e2e\u3059\u308b\u3002 1)\u6df1\u3044\u5208\u308a\u8fbc\u307f\u3068(2)\u96a0\u308c\u305f\/\u4fdd\u6301\/MLP(\u5e45)\u5208\u308a\u8fbc\u307f\u3068\u3044\u30462\u3064\u306e\u7570\u306a\u308b\u5208\u308a\u51fa\u3057\u65b9\u3092\u63a2\u308b\u3002 \u3053\u306e\u30a2\u30d7\u30ed\u30fc\u30c1\u306f\u3001Llama 3.1 8B\u304b\u3089\u9b45\u529b\u7684\u306a4B\u30e2\u30c7\u30eb\u3001Mistral NeMo 12B\u304b\u3089\u6700\u5148\u7aef\u306eMistral-NeMo-Minitron-8B\u30e2\u30c7\u30eb\u3092\u751f\u6210\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2408.11796v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2408.11796v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Wed, 21 Aug 2024 17:38:48 GMT)<\/li>\n\n\n\n<li>MInitron\u30a2\u30d7\u30ed\u30fc\u30c1\uff08<a href=\"https:\/\/devneko.jp\/wordpress\/?p=5244\">MINITRON \/ Compact Language Models via Pruning and Knowledge Distillation \u2013 arXiv\u6700\u65b0\u8ad6\u6587\u306e\u7d39\u4ecb (devneko.jp)<\/a>\uff09\u3067\u306e\u516c\u958b\u30e2\u30c7\u30eb\u306e\u5c0f\u898f\u6a21\u5316<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/nvidia\/Mistral-NeMo-Minitron-8B-Base\">nvidia\/Mistral-NeMo-Minitron-8B-Base \u00b7 Hugging Face<\/a>\u3001<a href=\"https:\/\/huggingface.co\/nvidia\/Mistral-NeMo-Minitron-8B-Base\">nvidia\/Mistral-NeMo-Minitron-8B-Base \u00b7 Hugging Face<\/a>\u3001<a href=\"https:\/\/huggingface.co\/nvidia\/Mistral-NeMo-Minitron-8B-Base\">nvidia\/Mistral-NeMo-Minitron-8B-Base \u00b7 Hugging Face<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>xGen-MM (BLIP-3): A Family of Open Large Multimodal Models\u00a0<\/strong>[157.4]<br>\u672c\u7a3f\u3067\u306f,LMM(Large Multimodal Models)\u3092\u958b\u767a\u3059\u308b\u305f\u3081\u306e\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308bxGen-MM\u3092\u7d39\u4ecb\u3059\u308b\u3002 \u3053\u306e\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306f\u3001\u614e\u91cd\u306b\u30ad\u30e5\u30ec\u30fc\u30c8\u3055\u308c\u305f\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30ec\u30b7\u30d4\u3001\u30e2\u30c7\u30eb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3001\u7d50\u679c\u306eLMM\u30b9\u30a4\u30fc\u30c8\u3067\u69cb\u6210\u3055\u308c\u3066\u3044\u308b\u3002 \u79c1\u305f\u3061\u306e\u30e2\u30c7\u30eb\u306f\u3001\u30b7\u30f3\u30b0\u30eb\u30a4\u30e1\u30fc\u30b8\u3068\u30de\u30eb\u30c1\u30a4\u30e1\u30fc\u30b8\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3092\u542b\u3080\u3001\u3055\u307e\u3056\u307e\u306a\u30bf\u30b9\u30af\u306b\u308f\u305f\u3063\u3066\u53b3\u683c\u306a\u8a55\u4fa1\u3092\u53d7\u3051\u3066\u3044\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2408.08872v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2408.08872v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Fri, 16 Aug 2024 17:57:01 GMT)<\/li>\n\n\n\n<li>xGen-MM (also known as BLIP-3), a framework for developing Large Multimodal Models (LMMs)<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f\u3001<a href=\"https:\/\/github.com\/salesforce\/LAVIS\/tree\/xgen-mm\">GitHub &#8211; salesforce\/LAVIS at xgen-mm<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u5148\u9031\u3082\u516c\u958b\u30e2\u30c7\u30eb\u306b\u95a2\u3059\u308b\u5831\u544a\u304c\u51fa\u3066\u3044\u305f\u3002HERMES 3\u306fHERMES\u30b7\u30ea\u30fc\u30ba\u306e\u6700\u65b0\u7248\u3067Llama3.1\u30d9\u30fc\u30b9(Paper page &#8211; Hermes 3 Technical Report (huggingf &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=5330\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;HERMES 3, Jamba 1.5, Mistral-NeMo-Minitron-8B, Llama-3.1-Minitron-4B, BLIP-3&#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":[223,293],"class_list":["post-5330","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-oss"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5330","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=5330"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5330\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5330"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5330"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5330"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}