{"id":3830,"date":"2023-09-18T07:51:00","date_gmt":"2023-09-17T22:51:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3830"},"modified":"2023-09-18T07:51:00","modified_gmt":"2023-09-17T22:51:00","slug":"are-large-language-model-based-evaluators-the-solution-to-scaling-up-multilingual-evaluation","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3830","title":{"rendered":"Are Large Language Model-based Evaluators the Solution to Scaling Up Multilingual Evaluation?"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Are Large Language Model-based Evaluators the Solution to Scaling Up Multilingual Evaluation?\u00a0<\/strong>[20.5]<br>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u306f\u3001\u81ea\u7136\u8a00\u8a9e\u51e6\u7406(NLP)\u30bf\u30b9\u30af\u306b\u304a\u3044\u3066\u7d20\u6674\u3089\u3057\u3044\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u793a\u3057\u3066\u3044\u308b\u3002 \u73fe\u5728\u306e\u8a55\u4fa1\u6280\u8853\u3067\u306f\u3001\u9069\u5207\u306a\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3001\u30e1\u30c8\u30ea\u30af\u30b9\u3001\u30b3\u30b9\u30c8\u3001\u4eba\u9593\u306e\u30a2\u30ce\u30c6\u30fc\u30b7\u30e7\u30f3\u3078\u306e\u30a2\u30af\u30bb\u30b9\u304c\u6b20\u5982\u3057\u3066\u3044\u308b\u3002 \u672c\u7a3f\u3067\u306f,LLM\u306b\u57fa\u3065\u304f\u8a55\u4fa1\u5668\u304c\u591a\u8a00\u8a9e\u8a55\u4fa1\u306e\u30b9\u30b1\u30fc\u30eb\u30a2\u30c3\u30d7\u306b\u6709\u52b9\u304b\u3069\u3046\u304b\u3092\u691c\u8a0e\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2309.07462v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2309.07462v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 14 Sep 2023 06:41:58 GMT)<\/li>\n\n\n\n<li>LLM\u304cNLP\u306e\u8a55\u4fa1\u5668\u3068\u3057\u3066\u591a\u8a00\u8a9e\u8a2d\u5b9a\u3067\u3046\u307e\u304f\u3044\u304f\u304b\u8a55\u4fa1\u3057\u305f\u8ad6\u6587\u3002\u300cWe see that the PA between the annotators and GPT is lowest compared to the PA between the human annotators for Japanese and Czech\u300d\uff08PA: Percentage Agreement\u00a0\uff09\u300cOur work indicates that LLMbased evaluators need to be used cautiously in the multilingual setting, particularly on languages on which LLMs are known to perform poorly.\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li>GPT-4\u3068\u304b\u3060\u3068\u82f1\u8a9e\u3067\u6709\u52b9\u3060\u3063\u305f\u624b\u6cd5\u304c\u65e5\u672c\u8a9e\u3067\u3082\u52d5\u304f\uff08\u3088\u3046\u306b\u898b\u3048\u308b\uff09\u308f\u3051\u3060\u304c\u3001\u6b63\u3057\u304f\u52d5\u4f5c\u3057\u3066\u3044\u308b\u304b\u3069\u3046\u304b\u691c\u8a3c\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u3001\u3068\u3044\u3046\u5f53\u7136\u3068\u8a00\u3048\u3070\u5f53\u7136\u306e\u7d50\u679c\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":[223,224,267,647],"class_list":["post-3830","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-llm-as-a-judge","tag-multilingual","tag-647"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3830","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=3830"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3830\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3830"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3830"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3830"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}