{"id":1562,"date":"2022-02-28T06:00:00","date_gmt":"2022-02-27T21:00:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=1562"},"modified":"2022-02-28T06:00:00","modified_gmt":"2022-02-27T21:00:00","slug":"sparse-expert-model%e3%81%ae%e5%8a%b9%e7%8e%87%e7%9a%84%e8%a8%ad%e8%a8%88","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=1562","title":{"rendered":"Sparse Expert Model\u306e\u52b9\u7387\u7684\u8a2d\u8a08"},"content":{"rendered":"\n<ul class=\"wp-block-list\"><li><strong>Designing Effective Sparse Expert Models\u00a0<\/strong>[45.2]<br>MoE(Mixture-of-Experts)\u3068Switch Transformer\u306f\u3001\u3088\u308a\u5927\u304d\u304f\u3001\u3088\u308a\u6709\u80fd\u306a\u8a00\u8a9e\u30e2\u30c7\u30eb\u3078\u306e\u30a8\u30cd\u30eb\u30ae\u30fc\u52b9\u7387\u306e\u9ad8\u3044\u7d4c\u8def\u3068\u3057\u3066\u63d0\u6848\u3055\u308c\u3066\u3044\u308b\u3002 \u3057\u304b\u3057\u3001\u3055\u307e\u3056\u307e\u306a\u81ea\u7136\u8a00\u8a9e\u30bf\u30b9\u30af\u306e\u6700\u5148\u7aef\u5316\u306f\u3001\u5fae\u8abf\u6574\u4e2d\u306b\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u4e0d\u5b89\u5b9a\u3055\u3068\u4e0d\u78ba\u5b9f\u306a\u54c1\u8cea\u306b\u3088\u3063\u3066\u59a8\u3052\u3089\u308c\u3066\u3044\u308b\u3002\u00a0\u79c1\u305f\u3061\u306e\u4ed5\u4e8b\u306f\u3053\u308c\u3089\u306e\u554f\u984c\u306b\u7126\u70b9\u3092\u5f53\u3066\u3001\u30c7\u30b6\u30a4\u30f3\u30ac\u30a4\u30c9\u3068\u3057\u3066\u6a5f\u80fd\u3057\u3001\u63a8\u8ad6(SuperGLUE, ARC Easy, ARC Challenge)\u3001\u8981\u7d04(XSum, CNN-DM)\u3001\u30af\u30ed\u30fc\u30ba\u30c9\u30d6\u30c3\u30af\u8cea\u554f\u5fdc\u7b54(WebQA, Natural Questions)\u3001\u6575\u5bfe\u7684\u306b\u69cb\u7bc9\u3055\u308c\u305f\u30bf\u30b9\u30af(Winogrande, ANLI R3)\u306a\u3069\u3001\u3055\u307e\u3056\u307e\u306a\u30bf\u30b9\u30af\u306e\u96c6\u5408\u306b\u304a\u3044\u3066\u3001\u30c8\u30e9\u30f3\u30b9\u30d5\u30a1\u30fc\u30e9\u30fc\u30cb\u30f3\u30b0\u306b\u304a\u3051\u308b\u6700\u5148\u7aef\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u521d\u3081\u3066\u9054\u6210\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2202.08906v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2202.08906v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 17 Feb 2022 21:39:10 GMT)<ul><li>\u5927\u898f\u6a21\u30b9\u30d1\u30fc\u30b9\u30e2\u30c7\u30eb\u306e\u4e0d\u5b89\u5b9a\u3055\u3092\u89e3\u6d88\u3001\u69d8\u3005\u306a\u30bf\u30b9\u30af\u3067SoTA\u3068\u306e\u3053\u3068\u3002<\/li><li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/tensorflow\/mesh\">GitHub &#8211; tensorflow\/mesh: Mesh TensorFlow: Model Parallelism Made Easier<\/a><\/li><\/ul><\/li><\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Designing Effective Sparse Expert Models\u00a0[45.2]MoE(Mixture-of-Experts)\u3068Switch Transformer\u306f\u3001\u3088\u308a\u5927\u304d\u304f\u3001\u3088\u308a\u6709\u80fd\u306a\u8a00\u8a9e\u30e2\u30c7\u30eb\u3078\u306e\u30a8 &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=1562\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;Sparse Expert Model\u306e\u52b9\u7387\u7684\u8a2d\u8a08&#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":[250,278],"class_list":["post-1562","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-mixture-of-experts","tag-nlp"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/1562","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=1562"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/1562\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1562"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1562"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1562"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}