{"id":2681,"date":"2022-12-12T06:41:00","date_gmt":"2022-12-11T21:41:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=2681"},"modified":"2022-12-12T06:41:00","modified_gmt":"2022-12-11T21:41:00","slug":"deepspeed-data-efficiency","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=2681","title":{"rendered":"DeepSpeed Data Efficiency"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>DeepSpeed Data Efficiency: Improving Deep Learning Model Quality and Training Efficiency via Efficient Data Sampling and Routing\u00a0<\/strong>[31.7]<br>DeepSpeed Data Efficiency\u306f\u3001\u30c7\u30fc\u30bf\u306e\u5229\u7528\u6027\u3092\u5411\u4e0a\u3057\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u52b9\u7387\u3092\u5411\u4e0a\u3057\u3001\u30e2\u30c7\u30eb\u54c1\u8cea\u3092\u6539\u5584\u3059\u308b\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308b\u3002 \u30ab\u30ea\u30ad\u30e5\u30e9\u30e0\u5b66\u7fd2\u306b\u3088\u308b\u52b9\u7387\u7684\u306a\u30c7\u30fc\u30bf\u30b5\u30f3\u30d7\u30ea\u30f3\u30b0\u3068\u3001\u30e9\u30f3\u30c0\u30e0\u306a\u30ec\u30a4\u30e4\u30ef\u30a4\u30c9\u30c8\u30fc\u30af\u30f3\u30c9\u30ed\u30c3\u30d7\u306b\u3088\u308b\u52b9\u7387\u7684\u306a\u30c7\u30fc\u30bf\u30eb\u30fc\u30c6\u30a3\u30f3\u30b0\u3092\u63d0\u4f9b\u3059\u308b\u3002 GPT-3 1.3B \u3068 BERT-Large \u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u4e8b\u524d\u5b66\u7fd2\u306b\u30bd\u30ea\u30e5\u30fc\u30b7\u30e7\u30f3\u3092\u9069\u7528\u3059\u308b\u3053\u3068\u3067\u30011\/2\u306e\u30c7\u30fc\u30bf\u30681\/2\u306e\u6642\u9593\u3067\u540c\u69d8\u306e\u30e2\u30c7\u30eb\u54c1\u8cea\u3092\u5b9f\u73fe\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2212.03597v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2212.03597v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 7 Dec 2022 12:27:28 GMT)<\/li>\n\n\n\n<li>\u30c7\u30fc\u30bf\u7684\u306b\u3082\u8a08\u7b97\u8cc7\u6e90\u7684\u306b\u3082\u52b9\u7387\u7684\u306a\u5b66\u7fd2\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306e\u63d0\u6848\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/microsoft\/DeepSpeed\">microsoft\/DeepSpeed: DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective. (github.com)<\/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":[551],"class_list":["post-2681","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-551"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/2681","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=2681"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/2681\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2681"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2681"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2681"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}