{"id":6926,"date":"2025-06-18T05:38:00","date_gmt":"2025-06-17T20:38:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6926"},"modified":"2025-06-14T22:59:33","modified_gmt":"2025-06-14T13:59:33","slug":"domain2vec-vectorizing-datasets-to-find-the-optimal-data-mixture-without-training","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6926","title":{"rendered":"Domain2Vec: Vectorizing Datasets to Find the Optimal Data Mixture without Training\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Domain2Vec: Vectorizing Datasets to Find the Optimal Data Mixture without Training\u00a0<\/strong>[53.1]<br>DOMAIN2VEC\u306f\u3001\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u8907\u6570\u306e\u30e1\u30bf\u30c9\u30e1\u30a4\u30f3\u306e\u7dda\u5f62\u7d50\u5408\u306b\u5206\u89e3\u3059\u308b\u65b0\u3057\u3044\u30a2\u30d7\u30ed\u30fc\u30c1\u3067\u3059\u3002\u3053\u306e\u624b\u6cd5\u306f\u3001\u30c9\u30e1\u30a4\u30f3\u30d9\u30af\u30bf\u30fc\u3092\u751f\u6210\u3057\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306a\u3057\u3067\u30c7\u30fc\u30bf\u30df\u30af\u30b9\u30c1\u30e3\u30fc\u306e\u6700\u9069\u5316\u3092\u53ef\u80fd\u306b\u3057\u307e\u3059\u3002\u5b9f\u9a13\u3067\u306f\u3001\u3053\u306e\u65b9\u6cd5\u304c\u8a08\u7b97\u30b3\u30b9\u30c8\u3092\u6291\u3048\u306a\u304c\u3089\u3001\u4e0b\u6d41\u30bf\u30b9\u30af\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u5e73\u57472.83%\u5411\u4e0a\u3055\u305b\u308b\u3053\u3068\u304c\u793a\u3055\u308c\u3066\u3044\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2506.10952v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2506.10952v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 12 Jun 2025 17:53:51 GMT)<\/li>\n\n\n\n<li>\u8272\u3005\u306a\u52d5\u304d\u304c\u3042\u3063\u3066\u8208\u5473\u6df1\u30442vec\u7cfb\u306e\u5831\u544a<\/li>\n\n\n\n<li>\u300cDOMAIN2VEC seamlessly integrates with existing methods, greatly improving their efficiency and scalability by establishing a direct relationship between model performance and domain vectors, without requiring retraining when training datasets change. Our experimental results demonstrate that both DOMAIN2VEC+DA2 and DOMAIN2VEC+RegMix achieve comparable text generation and downstream task performance with reduced computational overhead com- pared to existing approaches.\u300d<\/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":[4],"class_list":["post-6926","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-2vec"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6926","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=6926"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6926\/revisions"}],"predecessor-version":[{"id":6927,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6926\/revisions\/6927"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6926"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6926"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6926"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}