{"id":2863,"date":"2023-02-02T06:31:00","date_gmt":"2023-02-01T21:31:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=2863"},"modified":"2023-02-02T06:31:00","modified_gmt":"2023-02-01T21:31:00","slug":"xlm-v-overcoming-the-vocabulary-bottleneck-in-multilingual-masked-language-models","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=2863","title":{"rendered":"XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models&nbsp;<\/strong>[87.7]<br>\u6211\u3005\u306f,\u8a00\u8a9e\u9593\u306e\u30c8\u30fc\u30af\u30f3\u5171\u6709\u3092\u975e\u5f37\u8abf\u306b\u3059\u308b\u3053\u3068\u3067,\u975e\u5e38\u306b\u5927\u304d\u306a\u591a\u8a00\u8a9e\u8a9e\u5f59\u306b\u62e1\u5f35\u3059\u308b\u65b0\u305f\u306a\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u63d0\u6848\u3059\u308b\u3002 \u6211\u3005\u306f100\u4e07\u306e\u30c8\u30fc\u30af\u30f3\u8a9e\u5f59\u3092\u6301\u3064\u591a\u8a00\u8a9e\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u3042\u308bXLM-V\u3092\u8a13\u7df4\u3059\u308b\u3002 XLM-V\u306f\u3001\u81ea\u7136\u8a00\u8a9e\u63a8\u8ad6(XNLI)\u304b\u3089\u8cea\u554f\u5fdc\u7b54(MLQA)\u3001\u540d\u524d\u4ed8\u304d\u30a8\u30f3\u30c6\u30a3\u30c6\u30a3\u8a8d\u8b58(WikiAnn)\u307e\u3067\u5e45\u5e83\u3044\u30bf\u30b9\u30af\u3067XLM-R\u3088\u308a\u512a\u308c\u3066\u3044\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2301.10472v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2301.10472v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Wed, 25 Jan 2023 09:15:17 GMT)<\/li>\n\n\n\n<li>\u591a\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u554f\u984c\uff08\u304b\u3064\u8b70\u8ad6\uff09\u3068\u306a\u308b\u8a9e\u5f59\u306b\u95a2\u3059\u308b\u8ad6\u6587\u3002100\u4e07\u8a9e\u5f59\u3092\u6301\u3064\u30e2\u30c7\u30eb\u3092\u5b66\u7fd2\u3057\u512a\u308c\u305f\u6027\u80fd\u3092\u51fa\u3057\u3066\u3044\u308b\u306e\u306f\u3059\u3054\u3044\u3002<\/li>\n\n\n\n<li>\u300cMost notably, we provide evidence showing that expanding the vocabulary beyond 1M tokens can degrade performance on downstream tasks\u300d\u3068\u3044\u3046\u306e\u3082\u8208\u5473\u6df1\u3044\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":[223,439,567],"class_list":["post-2863","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-vocabulary","tag-567"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/2863","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=2863"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/2863\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2863"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2863"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2863"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}