{"id":3861,"date":"2023-09-29T05:00:00","date_gmt":"2023-09-28T20:00:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3861"},"modified":"2023-09-29T05:00:00","modified_gmt":"2023-09-28T20:00:00","slug":"slimpajama-dc-understanding-data-combinations-for-llm-training","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3861","title":{"rendered":"SlimPajama-DC: Understanding Data Combinations for LLM Training"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>SlimPajama-DC: Understanding Data Combinations for LLM Training&nbsp;<\/strong>[35.3]<br>\u672c\u7a3f\u3067\u306f,SlimPajama\u3092\u7528\u3044\u305f\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u5b66\u7fd2\u306b\u304a\u3051\u308b\u5404\u7a2e\u30c7\u30fc\u30bf\u306e\u7d44\u307f\u5408\u308f\u305b\u306e\u5f71\u97ff\u3092\u7406\u89e3\u3059\u308b\u3053\u3068\u3092\u76ee\u7684\u3068\u3059\u308b\u3002 SlimPajama\u306f\u53b3\u683c\u306b\u91cd\u8907\u3057\u305f\u30de\u30eb\u30c1\u30bd\u30fc\u30b9\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3067\u3001627B\u30c8\u30fc\u30af\u30f3\u306b\u3055\u3089\u306b\u91cd\u8907\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2309.10818v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2309.10818v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Tue, 19 Sep 2023 17:59:54 GMT)<\/li>\n\n\n\n<li>LLM\u5b66\u7fd2\u6642\u306e\u30c7\u30fc\u30bf\u3092\u3069\u3046\u7d44\u307f\u5408\u308f\u305b\u308b\u304b\u691c\u8a3c\u3057\u305f\u5831\u544a\u3001\u300ca lower training loss doesn\u2019t necessarily correlate directly with superior model performance.\u300d\u300cThis implies that as the amount of code in training increases, the training loss diminishes.\u300d\u3068\u3044\u3046\u306e\u304c\u9762\u767d\u3044\uff08\u8a00\u308f\u308c\u3066\u307f\u308c\u3070\u307e\u3041\u305d\u3046\u3060\u308d\u3046\u3068\u601d\u3046\u3002\u3002\uff09<\/li>\n\n\n\n<li>\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306f<a href=\"https:\/\/huggingface.co\/datasets\/cerebras\/SlimPajama-627B\">cerebras\/SlimPajama-627B \u00b7 Datasets at Hugging Face<\/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":[95,223,491],"class_list":["post-3861","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-data-slimming","tag-llm","tag-491"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3861","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=3861"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3861\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3861"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3861"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3861"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}