{"id":3917,"date":"2023-10-13T05:38:00","date_gmt":"2023-10-12T20:38:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3917"},"modified":"2023-10-13T05:38:00","modified_gmt":"2023-10-12T20:38:00","slug":"think-before-you-speak-training-language-models-with-pause-tokens","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3917","title":{"rendered":"Think before you speak: Training Language Models With Pause Tokens"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Think before you speak: Training Language Models With Pause Tokens\u00a0<\/strong>[78.0]<br>\u8a00\u8a9e\u30e2\u30c7\u30eb\u306f\u3001\u5373\u5ea7\u306b\u9023\u7d9a\u3057\u3066\u4e00\u9023\u306e\u30c8\u30fc\u30af\u30f3\u3092\u751f\u6210\u3057\u3066\u5fdc\u7b54\u3092\u751f\u6210\u3059\u308b\u3002 \u4ee3\u308f\u308a\u306b\u3001$(K+1)th$\u30c8\u30fc\u30af\u30f3\u3092\u51fa\u529b\u3059\u308b\u524d\u306b\u3001\u30e2\u30c7\u30eb\u306b$K+10$\u96a0\u308c\u30d9\u30af\u30bf\u30fc\u3092\u64cd\u4f5c\u3055\u305b\u308b\u3068\u3057\u305f\u3089\u3069\u3046\u3067\u3057\u3087\u3046? \u79c1\u305f\u3061\u306f\u3001(\u5b66\u7fd2\u53ef\u80fd\u306a)$textitpause$\u30c8\u30fc\u30af\u30f3\u3092\u4f7f\u3063\u3066\u3001\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3068\u63a8\u8ad6\u3092\u884c\u3046\u3053\u3068\u3067\u3001\u3053\u306e\u30a2\u30a4\u30c7\u30a2\u3092\u904b\u7528\u3057\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2310.02226v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2310.02226v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 3 Oct 2023 17:32:41 GMT)<\/li>\n\n\n\n<li>\u00a0pause-inference (and pause-finetuning)\u3068\u3057\u3066\u63a8\u8ad6\uff08\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\uff09\u6642\u306b&lt;pause> token\u3092\u5c0e\u5165\u3001pause\u4e2d\u306f\u51fa\u529b\u3092\u6291\u5236\u3001\u6291\u5236\u3057\u3066\u3044\u308b\u5206\u3060\u3051\uff08&lt;pause> token\u5206\u3060\u3051\uff09\u8a08\u7b97\u7d4c\u8def\u3092\u5897\u52a0\u3001\u6027\u80fd\u304c\u5411\u4e0a\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":[223],"class_list":["post-3917","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3917","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=3917"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3917\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3917"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3917"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3917"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}