{"id":7030,"date":"2025-07-01T03:11:00","date_gmt":"2025-06-30T18:11:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7030"},"modified":"2025-06-29T10:15:13","modified_gmt":"2025-06-29T01:15:13","slug":"what-matters-in-llm-generated-data-diversity-and-its-effect-on-model-fine-tuning","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7030","title":{"rendered":"What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning\u00a0["},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning\u00a0<\/strong>[22.4]<br>LLM\u751f\u6210\u30c7\u30fc\u30bf\u306e\u591a\u69d8\u6027\u30ec\u30d9\u30eb\u304c\u4e0b\u6d41\u30e2\u30c7\u30eb\u306e\u6027\u80fd\u306b\u3069\u306e\u3088\u3046\u306b\u5f71\u97ff\u3059\u308b\u304b\u3092\u793a\u3059\u3002 \u307e\u305f\u3001LLM\u751f\u6210\u30c7\u30fc\u30bf\u306e\u7570\u306a\u308b\u5272\u5408\u3092\u6df7\u5408\u3057\u305f\u30c7\u30fc\u30bf\u306b\u57fa\u3065\u3044\u3066\u8a13\u7df4\u3055\u308c\u305f\u30e2\u30c7\u30eb\u306e\u6027\u80fd\u306b\u3064\u3044\u3066\u3082\u691c\u8a0e\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2506.19262v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2506.19262v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 24 Jun 2025 02:44:58 GMT)<\/li>\n\n\n\n<li>\u5408\u6210\u30c7\u30fc\u30bf\u304c\u4e0e\u3048\u308b\u5f71\u97ff\u306b\u95a2\u3059\u308b\u5831\u544a\u3002\u7279\u306b\u591a\u69d8\u6027\u306e\u5ea6\u5408\u3044\u306b\u6ce8\u76ee\u3057\u3066\u3044\u308b\u3002<\/li>\n\n\n\n<li>\u300cOur experimental results show that, with minimal distribution shift, moderately diverse LLM-generated data can enhance model performance in scenarios with insufficient labeled data, whereas highly diverse generated data has a negative impact.\u300d\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":[71,390],"class_list":["post-7030","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-collapse","tag-synthetic-data"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7030","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=7030"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7030\/revisions"}],"predecessor-version":[{"id":7031,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7030\/revisions\/7031"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7030"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7030"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7030"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}