{"id":4535,"date":"2024-03-15T06:04:00","date_gmt":"2024-03-14T21:04:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4535"},"modified":"2024-03-15T06:04:00","modified_gmt":"2024-03-14T21:04:00","slug":"galore-low-rank-projection","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4535","title":{"rendered":"GaLore: Low-Rank Projection"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection\u00a0<\/strong>[139.2]<br>LLM(Large Language Models)\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306f\u3001\u91cd\u307f\u4ed8\u3051\u3084GPU\u72b6\u614b\u306e\u5897\u5927\u306b\u3088\u3063\u3066\u3001\u30e1\u30e2\u30ea\u4e0a\u306e\u91cd\u5927\u306a\u554f\u984c\u304c\u767a\u751f\u3059\u308b\u3002 \u672c\u7814\u7a76\u3067\u306f,\u30e1\u30e2\u30ea\u52b9\u7387\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6226\u7565\u3068\u3057\u3066\u30b0\u30e9\u30c7\u30a3\u30a8\u30f3\u30c8\u30fb\u30ed\u30fc\u30e9\u30f3\u30c9\u30fb\u30d7\u30ed\u30b8\u30a7\u30af\u30b7\u30e7\u30f3(GaLore)\u3092\u63d0\u6848\u3059\u308b\u3002 \u79c1\u305f\u3061\u306e8\u30d3\u30c3\u30c8\u306eGaLore\u306f\u3001BF16\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u3068\u6bd4\u8f03\u3057\u3066\u3001\u30e1\u30e2\u30ea\u309282.5%\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u7dcf\u30e1\u30e2\u30ea\u309263.3%\u524a\u6e1b\u3057\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2403.03507v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2403.03507v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 6 Mar 2024 07:29:57 GMT)<\/li>\n\n\n\n<li>LLM\u3092\u6271\u3046\u4e0a\u3067\u5927\u554f\u984c\u306b\u306a\u308b\u30e1\u30e2\u30ea\u52b9\u7387\u3092\u9ad8\u3081\u305f\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u624b\u6cd5\u306e\u63d0\u6848\u3002NVIDIA RTX 4090 RAM 24GB\u30677B\u30e2\u30c7\u30eb\u3092\u4e8b\u524d\u5b66\u7fd2\u53ef\u80fd\u3068\u306e\u3053\u3068\u3002<\/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-4535","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\/4535","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=4535"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4535\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4535"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4535"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4535"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}