{"id":4726,"date":"2024-04-15T05:05:00","date_gmt":"2024-04-14T20:05:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4726"},"modified":"2024-04-15T05:05:00","modified_gmt":"2024-04-14T20:05:00","slug":"rho-1-not-all-tokens-are-what-you-need","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4726","title":{"rendered":"Rho-1: Not All Tokens Are What You Need"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Rho-1: Not All Tokens Are What You Need\u00a0<\/strong>[132.3]<br>\u300c\u30b3\u30fc\u30d1\u30b9\u5185\u306e\u30c8\u30fc\u30af\u30f3\u306f\u3059\u3079\u3066\u3001\u8a00\u8a9e\u30e2\u30c7\u30eb\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306b\u7b49\u3057\u304f\u91cd\u8981\u3067\u306f\u306a\u3044\u300d Rho-1 \u306f\u9078\u629e\u8a00\u8a9e\u30e2\u30c7\u30ea\u30f3\u30b0 (SLM) \u3092\u63a1\u7528\u3057\u3066\u304a\u308a\u3001\u6240\u671b\u306e\u5206\u5e03\u306b\u5408\u308f\u305b\u3066\u6709\u7528\u306a\u30c8\u30fc\u30af\u30f3\u3092\u9078\u629e\u7684\u306b\u8a13\u7df4\u3059\u308b\u3002 15B OpenWebMath\u30b3\u30fc\u30d1\u30b9\u3067\u7d99\u7d9a\u4e8b\u524d\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3092\u884c\u3046\u3068\u3001Rho-1\u306f9\u3064\u306e\u6570\u5b66\u30bf\u30b9\u30af\u3067\u6700\u592730%\u306e\u30b7\u30e7\u30c3\u30c8\u7cbe\u5ea6\u3067\u7d76\u5bfe\u7684\u306b\u6539\u5584\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2404.07965v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2404.07965v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 11 Apr 2024 17:52:01 GMT)<\/li>\n\n\n\n<li>\u300cSelective Language Modeling (SLM), which selectively trains on useful tokens that aligned with the desired distribution.\u300d\u306b\u3088\u3063\u3066\u6700\u7d42\u6027\u80fd\u304c\u4e0a\u304c\u308b\u3068\u3044\u3046\u5831\u544a\u3002\u9ad8\u54c1\u8cea\uff08\u6240\u671b\u306e\uff09\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u3067\u53c2\u7167\u30e2\u30c7\u30eb\u3092\u69cb\u7bc9\u3057\u3001\u305d\u306e\u7d50\u679c\u3092\u5229\u7528\u3057\u3066\u30c8\u30fc\u30af\u30f3\u3092\u9078\u629e\u3059\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u306e\u3088\u3046\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/microsoft\/rho\">microsoft\/rho: Token-level Data Filtering &amp; Selective Pretraining of LLMs. (github.com)<\/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":[223,551],"class_list":["post-4726","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-551"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4726","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=4726"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4726\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4726"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4726"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4726"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}