{"id":814,"date":"2021-10-13T05:00:00","date_gmt":"2021-10-12T20:00:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=814"},"modified":"2021-10-13T05:00:00","modified_gmt":"2021-10-12T20:00:00","slug":"m6-10t-megatron-turing-nlg-530b-%e5%b7%a8%e5%a4%a7%e3%81%aa%e4%ba%8b%e5%89%8d%e5%ad%a6%e7%bf%92%e3%83%a2%e3%83%87%e3%83%ab","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=814","title":{"rendered":"M6-10T \/ Yuan 1.0 \/ Megatron-Turing NLG 530B \u5de8\u5927\u306a\u4e8b\u524d\u5b66\u7fd2\u30e2\u30c7\u30eb"},"content":{"rendered":"\n<ul class=\"wp-block-list\"><li><strong>M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining&nbsp;<\/strong>[55.2]<br>\u6975\u7aef\u306a\u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306b\u306f\u5927\u91cf\u306e\u8a08\u7b97\u3068\u30e1\u30e2\u30ea\u30d5\u30c3\u30c8\u30d7\u30ea\u30f3\u30c8\u304c\u5fc5\u8981\u3067\u3059\u3002 \u672c\u7a3f\u3067\u306f,\u9ad8\u30e1\u30e2\u30ea\u30d5\u30c3\u30c8\u30d7\u30ea\u30f3\u30c8\u8981\u6c42\u5927\u30e2\u30c7\u30eb\u306e\u305f\u3081\u306e\u7c21\u5358\u306a\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6226\u7565&#8221;Pseudo-to-Real&#8221;\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2110.03888v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2110.03888v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper\/translated\/2110.03888.pdf.html\">\u53c2\u8003\u8a33\uff08\u5168\u6587\uff09<\/a>&nbsp;&nbsp;(Fri, 8 Oct 2021 04:24:51 GMT)<ul><li>\u6975\u3081\u3066\u5de8\u5927\u306a\u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u5b66\u7fd2\u65b9\u6cd5\u3092\u6271\u3063\u305f\u8ad6\u6587\u300210\u65e5\u4ee5\u5185\u306b512\u500b\u306eGPU\u3060\u3051\u3067\u3001\u975e\u5e38\u306b\u5de8\u5927\u306a10T\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u30e2\u30c7\u30eb\u3092\u4e8b\u524d\u5b66\u7fd2\u53ef\u80fd\u3068\u306e\u3053\u3068\u3002<\/li><\/ul><\/li><\/ul>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>Yuan 1.0: Large-Scale Pre-trained Language Model in Zero-Shot and Few-Shot Learning&nbsp;<\/strong>[20.6]<br>GPT-3\u306e\u3088\u3046\u306a\u6700\u8fd1\u306e\u7814\u7a76\u306f\u3001\u591a\u304f\u306e\u81ea\u7136\u8a00\u8a9e\u51e6\u7406(NLP)\u30bf\u30b9\u30af\u306b\u304a\u3051\u308bZero-Shot\u3068Few-Shot\u5b66\u7fd2\u306e\u512a\u308c\u305f\u6027\u80fd\u3092\u793a\u3057\u3066\u3044\u308b\u3002 \u672c\u7a3f\u3067\u306f,\u5927\u898f\u6a21\u5206\u6563\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6027\u80fd\u3092\u30e2\u30c7\u30eb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u8a2d\u8a08\u306b\u7d44\u307f\u8fbc\u3080\u624b\u6cd5\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2110.04725v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2110.04725v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Sun, 10 Oct 2021 07:40:22 GMT)<ul><li>\u4e2d\u56fd\u306e245B\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u5de8\u5927\u306a\u30e2\u30c7\u30eb\u30025TB\u306e\u9ad8\u54c1\u8cea\u306a\u4e2d\u56fd\u8a9e\u306e\u30c7\u30fc\u30bf\u3067\u5b66\u7fd2\uff084095 PetaFlops-days\u3002\u3002\u3002\uff09\u3057\u69d8\u3005\u306a\u30bf\u30b9\u30af\u3067\u512a\u308c\u305f\u6027\u80fd\u3092\u78ba\u8a8d\u3002<\/li><\/ul><\/li><\/ul>\n\n\n\n<ul class=\"wp-block-list\"><li>NVIDIA\u3068Microsoft\u304c\u300cMegatron-Turing NLG 530B\u300d\u3068\u3044\u3046530B\u30d1\u30e9\u30e1\u30fc\u30bf\uff08GPT-3\u306e3\u500d\uff09\u306e\u30e2\u30c7\u30eb\u3092\u767a\u8868\u3057\u3066\u3044\u308b\u3002\uff08<a href=\"https:\/\/developer.nvidia.com\/blog\/using-deepspeed-and-megatron-to-train-megatron-turing-nlg-530b-the-worlds-largest-and-most-powerful-generative-language-model\/\">NVIDIA\u306e\u8a18\u4e8b<\/a>\u3001<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/blog\/using-deepspeed-and-megatron-to-train-megatron-turing-nlg-530b-the-worlds-largest-and-most-powerful-generative-language-model\/\">Microsoft\u306e\u8a18\u4e8b<\/a>\uff09zero-shot\u3001few-shot\u305d\u308c\u305e\u308c\u3067\u6027\u80fd\u304c\u6539\u5584\u3057\u3066\u304a\u308a\u5de8\u5927\u306a\u30e2\u30c7\u30eb\u69cb\u7bc9\u7af6\u4e89\u306f\u3057\u3070\u3089\u304f\u7d9a\u304d\u305d\u3046\u3002<ul><li>\u8a18\u4e8b\u306b\u3042\u308b\u300cWe live in a time where AI advancements are far outpacing Moore\u2019s law.\u300d\u3068\u3044\u3046\u8a18\u8f09\u304c\u8208\u5473\u6df1\u3044\u3002<\/li><\/ul><\/li><\/ul>\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining&nbsp;[55.2]\u6975\u7aef\u306a\u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=814\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;M6-10T \/ Yuan 1.0 \/ Megatron-Turing NLG 530B \u5de8\u5927\u306a\u4e8b\u524d\u5b66\u7fd2\u30e2\u30c7\u30eb&#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":[170,567],"class_list":["post-814","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-gigantic-language-model","tag-567"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/814","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=814"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/814\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=814"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=814"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=814"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}