{"id":5313,"date":"2024-08-22T04:53:00","date_gmt":"2024-08-21T19:53:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5313"},"modified":"2024-08-22T04:53:00","modified_gmt":"2024-08-21T19:53:00","slug":"leveraging-web-crawled-data-for-high-quality-fine-tuning","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5313","title":{"rendered":"Leveraging Web-Crawled Data for High-Quality Fine-Tuning"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Leveraging Web-Crawled Data for High-Quality Fine-Tuning\u00a0<\/strong>[24.2]<br>\u6211\u3005\u306f\u3001GPT-4\u306e\u3088\u3046\u306a\u5148\u9032\u7684\u306a\u30e2\u30c7\u30eb\u306b\u983c\u308b\u3053\u3068\u306a\u304f\u3001\u9ad8\u54c1\u8cea\u306a\u6559\u5e2b\u4ed8\u304d\u5fae\u8abf\u6574\u306e\u305f\u3081\u306e\u8cb4\u91cd\u306a\u60c5\u5831\u6e90\u3068\u3057\u3066\u3001Webcrawled Data\u304c\u6709\u52b9\u3067\u3042\u308b\u3068\u4e3b\u5f35\u3057\u3066\u3044\u308b\u3002 \u6211\u3005\u306f\u3001Webcrawled\u30c7\u30fc\u30bf\u3092\u3088\u308a\u5c0f\u3055\u306a\u9ad8\u54c1\u8cea\u306a\u30c7\u30fc\u30bf\u96c6\u5408\u3068\u6574\u5217\u3055\u305b\u308b\u3053\u3068\u3067\u3001\u30da\u30a2\u5316\u3055\u308c\u305f\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u81ea\u52d5\u751f\u6210\u3059\u308b\u3002 \u5b9f\u9a13\u306e\u7d50\u679c, \u30e2\u30c7\u30eb\u5909\u63db\u30c7\u30fc\u30bf\u3092\u7528\u3044\u305f\u5b66\u7fd2\u306f, \u4e2d\u56fd\u306b\u304a\u3051\u308b\u6570\u5b66\u554f\u984c\u306e\u5e73\u5747\u30b9\u30b3\u30a29.4%\u3067, \u9ad8\u54c1\u8cea\u306a\u30c7\u30fc\u30bf\u306e\u307f\u306b\u3088\u308b\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3092\u4e0a\u56de\u308a, \u3088\u308a\u826f\u3044\u7d50\u679c\u304c\u5f97\u3089\u308c\u308b\u3053\u3068\u304c\u308f\u304b\u3063\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2408.08003v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2408.08003v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 15 Aug 2024 08:12:52 GMT)<\/li>\n\n\n\n<li>\u300cDrawing on the intuition that rewriting data is comparatively simpler than performing intricate reasoning tasks for LLMs, we propose a method to augment the dataset by converting web-crawled data into high-quality ones.\u300d\u3068\u3044\u3046\u624b\u6cd5\u306e\u63d0\u6848\u3002\u5c0f\u898f\u6a21\u306a\u30b7\u30fc\u30c9\u30c7\u30fc\u30bf\u3068\u5927\u898f\u6a21\u306a\u30af\u30ed\u30fc\u30ea\u30f3\u30b0\u30c7\u30fc\u30bf\u306e\u30de\u30c3\u30c1\u30f3\u30b0\u3092\u3068\u308a\u3001\u5c02\u7528\u30e2\u30c7\u30eb\u3092\u4f5c\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u3002\u30af\u30ea\u30fc\u30cb\u30f3\u30b0\u304c\u5927\u5909\u306a\u306e\u304c\u4f1d\u308f\u3063\u3066\u304f\u308b\u3002\u6570\u5b66\u7684\u306a\u554f\u984c\u306e\u30d0\u30ea\u30a8\u30fc\u30b7\u30e7\u30f3\u306f\u3068\u3066\u3082\u591a\u305d\u3046\u3060\u3051\u3069\u3001\u3053\u306e\u65b9\u91dd\u3067\u3046\u307e\u304f\u3044\u304f\u306e\u306f\u306a\u305c\u306a\u306e\u3060\u308d\u3046\u304b\u3002\u3002\u3002\uff08\u6709\u540d\u306a\u554f\u984c\u306e\u5225\u89e3\u60c5\u5831\u304c\u4f7f\u308f\u308c\u3066\u3044\u308b\u3093\u3060\u308d\u3046\u304b\uff09<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/zhouj8553\/Web_to_SFT\">GitHub &#8211; zhouj8553\/Web_to_SFT: official code for the paper &#8220;Leveraging Web-Crawled Data for High-Quality Fine-Tuning&#8221;<\/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":[92],"class_list":["post-5313","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-data-augmentation"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5313","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=5313"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5313\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5313"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5313"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5313"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}