{"id":3753,"date":"2023-09-07T05:01:00","date_gmt":"2023-09-06T20:01:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3753"},"modified":"2023-09-07T05:01:00","modified_gmt":"2023-09-06T20:01:00","slug":"the-belebele-benchmark","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3753","title":{"rendered":"The Belebele Benchmark"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants\u00a0<\/strong>[82.6]<br>\u79c1\u305f\u3061\u306f122\u306e\u8a00\u8a9e\u5909\u7a2e\u306b\u307e\u305f\u304c\u308b\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3067\u3042\u308bBelebele\u3092\u7d39\u4ecb\u3057\u307e\u3059\u3002 \u3053\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306f\u3001\u9ad8\u3001\u4e2d\u3001\u4f4e\u30ea\u30bd\u30fc\u30b9\u8a00\u8a9e\u306b\u304a\u3051\u308b\u30c6\u30ad\u30b9\u30c8\u30e2\u30c7\u30eb\u306e\u8a55\u4fa1\u3092\u53ef\u80fd\u306b\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2308.16884v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2308.16884v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 31 Aug 2023 17:43:08 GMT)<\/li>\n\n\n\n<li>\u300cmultiple-choice machine reading comprehension (MRC) dataset spanning 122 language variants.\u300d\u3068\u3044\u3046\u3053\u3068\u3067\u975e\u5e38\u306b\u591a\u8a00\u8a9e\u306eMRC\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3002\u6a5f\u68b0\u7ffb\u8a33\u306b\u304a\u3051\u308bFLORES-200\u306e\u3088\u3046\u306a\u7acb\u3061\u4f4d\u7f6e\u3067\u975e\u5e38\u306b\u8cb4\u91cd\u306a\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8<\/li>\n\n\n\n<li>\u300cGPT3.5-TURBO performs the best on the top 20 languages, but after 40-50, its performance falls far behind INFOXLM and XLM-V.\u300d\u3068\u3044\u3046\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u7d50\u679c\u304c\u8208\u5473\u6df1\u3044\u3002\u5546\u696d\u30b7\u30b9\u30c6\u30e0\u306f\u3042\u308b\u7a0b\u5ea6\u30bf\u30fc\u30b2\u30c3\u30c8\u3068\u306a\u308b\u8a00\u8a9e\u3092\u7d5e\u3063\u3066\u3044\u308b\u3088\u3046\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/facebookresearch\/belebele\">GitHub &#8211; facebookresearch\/belebele: Repo for the Belebele dataset, a massively multilingual reading comprehension dataset.<\/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":[267,491],"class_list":["post-3753","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-multilingual","tag-491"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3753","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=3753"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3753\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3753"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3753"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3753"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}