{"id":4955,"date":"2024-06-03T04:30:00","date_gmt":"2024-06-02T19:30:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4955"},"modified":"2024-06-03T04:30:00","modified_gmt":"2024-06-02T19:30:00","slug":"yuan-2-0-m32-zamba-map-neo","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4955","title":{"rendered":"Yuan 2.0-M32, Zamba, MAP-Neo"},"content":{"rendered":"\n<p>\u4eca\u9031\u3082\u8208\u5473\u6df1\u3044LLM\u304c\u767a\u8868\u3055\u308c\u3066\u3044\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>MoE\u3067\u5c0f\u578b\u5f37\u529b\u306aYuan 2.0-M32<\/li>\n\n\n\n<li>SSM\uff08\uff06Transformer\u306e\u30cf\u30a4\u30d6\u30ea\u30c3\u30c9\uff09\u3067\u3042\u308b\u304c7B\u3068\u5b9f\u7528\u30b5\u30a4\u30ba\u304b\u3064Transformer\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e7B\u3068\u7af6\u5408\u3059\u308b\u6027\u80fd\u306b\u898b\u3048\u308bZamba<\/li>\n\n\n\n<li>\u4e2d\u56fd\u8a9e-\u82f1\u8a9e\u3067\u306f\u3042\u308b\u304c\u5f37\u529b\u306a\u30aa\u30fc\u30d7\u30f3\u30e2\u30c7\u30eb\u3067\u3042\u308bMAP-Neo<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Yuan 2.0-M32: Mixture of Experts with Attention Router\u00a0<\/strong>[30.9]<br>Yuan 2.0-M32\u306f\u3001Yuan-2.0 2B\u3068\u540c\u69d8\u306e\u30d9\u30fc\u30b9\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3067\u300132\u4eba\u306e\u30a8\u30ad\u30b9\u30d1\u30fc\u30c8\u30682\u4eba\u306e\u30a8\u30ad\u30b9\u30d1\u30fc\u30c8\u304c\u6d3b\u52d5\u3059\u308b\u6df7\u5408\u5c02\u9580\u5bb6\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u4f7f\u7528\u3057\u3066\u3044\u308b\u3002 \u65b0\u3057\u3044\u30eb\u30fc\u30bf\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3067\u3042\u308bAttention Router\u304c\u63d0\u6848\u3055\u308c\u3001\u3088\u308a\u52b9\u7387\u7684\u306a\u5c02\u9580\u5bb6\u306e\u9078\u629e\u306e\u305f\u3081\u306b\u63a1\u7528\u3055\u308c\u3001\u5f93\u6765\u306e\u30eb\u30fc\u30bf\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3068\u6bd4\u8f03\u3057\u30663.8%\u306e\u7cbe\u5ea6\u304c\u5411\u4e0a\u3059\u308b\u3002 Yuan 2.0-M32\u306f\u3001\u30b3\u30fc\u30c7\u30a3\u30f3\u30b0\u3001\u6570\u5b66\u3001\u304a\u3088\u3073\u69d8\u3005\u306a\u5c02\u9580\u5206\u91ce\u306b\u304a\u3051\u308b\u7af6\u4e89\u529b\u3092\u793a\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2405.17976v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2405.17976v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 28 May 2024 09:05:08 GMT)<\/li>\n\n\n\n<li>MoE\u3067\u30a2\u30af\u30c6\u30a3\u30d6\u30d1\u30e9\u30e1\u30fc\u30bf\u304c\u5c11\u306a\u3044\u304c\u512a\u308c\u305f\u6027\u80fd\u3092\u4e3b\u5f35\u3059\u308bLLM\u3002\u591a\u304f\u306e\u30bf\u30b9\u30af\u3067\u30a2\u30af\u30c6\u30a3\u30d6\u30d1\u30e9\u30e1\u30fc\u30bf\u7684\u306b\u540c\u898f\u6a21\u306ePhi-3\u3001\u500d\u4ee5\u4e0a\u306e\u898f\u6a21\u306eLlama-3 8B\u3088\u308a\u30b9\u30b3\u30a2\u304c\u9ad8\u3044\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/IEIT-Yuan\/Yuan2.0-M32\">GitHub &#8211; IEIT-Yuan\/Yuan2.0-M32: Mixture-of-Experts (MoE) Language Model<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Zamba: A Compact 7B SSM Hybrid Model\u00a0<\/strong>[11.0]<br>Zamba\u306f7B SSM\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u30cf\u30a4\u30d6\u30ea\u30c3\u30c9\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002 Zamba\u306f\u3001\u516c\u958b\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u304b\u30891T\u30c8\u30fc\u30af\u30f3\u3092\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u3002 Zamba\u306f\u3001\u540c\u7b49\u306e\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u30e2\u30c7\u30eb\u3088\u308a\u3082\u63a8\u8ad6\u304c\u304b\u306a\u308a\u901f\u3044\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2405.16712v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2405.16712v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Sun, 26 May 2024 22:23:02 GMT)<\/li>\n\n\n\n<li>SSM\u3068Transformer\u306e\u30cf\u30a4\u30d6\u30ea\u30c3\u30c9\u3067\u52b9\u7387\u7684\u3060\u304c\u5f37\u529b\u306aLLM<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/Zyphra\/Zamba-7B-v1\">Zyphra\/Zamba-7B-v1 \u00b7 Hugging Face<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>MAP-Neo: Highly Capable and Transparent Bilingual Large Language Model Series\u00a0<\/strong>[86.3]<br>\u79c1\u305f\u3061\u306fMAP-Neo\u3092\u30aa\u30fc\u30d7\u30f3\u30bd\u30fc\u30b9\u306b\u3057\u307e\u3057\u305f\u3002\u3053\u308c\u306f\u30014.5T\u306e\u9ad8\u54c1\u8cea\u30c8\u30fc\u30af\u30f3\u4e0a\u3067\u3001\u30b9\u30af\u30e9\u30c3\u30c1\u304b\u3089\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u305f7B\u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u6301\u3064\u30d0\u30a4\u30ea\u30f3\u30ac\u30eb\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u3059\u3002 <br><a href=\"http:\/\/arxiv.org\/abs\/2405.19327v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2405.19327v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 29 May 2024 17:57:16 GMT)<\/li>\n\n\n\n<li>\u5f37\u529b\u304b\u3064\u30aa\u30fc\u30d7\u30f3\u306aLLM<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/map-neo.github.io\/\">MAP-Neo<\/a>\u3001HuggingFace weight\u306f<a href=\"https:\/\/huggingface.co\/collections\/m-a-p\/neo-models-66395a5c9662bb58d5d70f04\">Neo-Models &#8211; a m-a-p Collection (huggingface.co)<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u4eca\u9031\u3082\u8208\u5473\u6df1\u3044LLM\u304c\u767a\u8868\u3055\u308c\u3066\u3044\u308b\u3002<\/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":[223,250,293,376],"class_list":["post-4955","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-mixture-of-experts","tag-oss","tag-ssm"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4955","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=4955"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4955\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4955"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4955"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4955"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}