{"id":3075,"date":"2023-03-23T05:17:00","date_gmt":"2023-03-22T20:17:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3075"},"modified":"2023-03-23T05:17:00","modified_gmt":"2023-03-22T20:17:00","slug":"consistency-analysis-of-chatgpt","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3075","title":{"rendered":"Consistency Analysis of ChatGPT"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Consistency Analysis of ChatGPT&nbsp;<\/strong>[54.2]<br>ChatGPT\u306f,\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\u306b\u57fa\u3065\u304f\u8cea\u554f\u30fb\u56de\u7b54\u5bfe\u8a71\u30b7\u30b9\u30c6\u30e0\u3067\u3042\u308b\u3002 \u8ad6\u7406\u7684\u306b\u4e00\u8cab\u3057\u305f\u884c\u52d5\u306b\u95a2\u3059\u308bChatGPT\u306e\u4fe1\u983c\u6027\u306b\u3064\u3044\u3066\u691c\u8a0e\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2303.06273v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2303.06273v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Sat, 11 Mar 2023 01:19:01 GMT)<\/li>\n\n\n\n<li>ChatGPT\u306e\u4e00\u8cab\u6027\u8a55\u4fa1\u3001<a href=\"https:\/\/aclanthology.org\/2022.coling-1.324.pdf\">BECEL: Benchmark for Consistency Evaluation of Language Models (aclanthology.org)<\/a>\u3092\u5229\u7528<\/li>\n\n\n\n<li>ChatGPT\u306fnegation consistency\u306b\u512a\u308c\u3066\u3044\u308b\u4e00\u65b9\u3067symmetric consistency\u304c\u3044\u307e\u3044\u3061\u3068\u3044\u3046\u7d50\u679c\u3002<\/li>\n\n\n\n<li>\u300cChatGPT fails to distinguish between the labels of \u201cNeutral\u201d and \u201cContradiction\u201d.\u300d\u3084\u300cMoreover, in inputorder invariant tasks, ChatGPT is likely to make a different decision when the order of the input sentences is switched.\u300d\u3068\u3044\u3046\u306e\u3082\u8208\u5473\u6df1\u3044\u3002\u8457\u8005\u3089\u306f\u30cf\u30a4\u30ea\u30b9\u30af\u306a\u7528\u9014\u306b\u4f7f\u3046\u306e\u306f\u81f4\u547d\u7684\u306a\u554f\u984c\u3068\u6307\u6458\u3057\u3066\u3044\u308b\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":[63,173],"class_list":["post-3075","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-chatgpt","tag-gpt"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3075","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=3075"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3075\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3075"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3075"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3075"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}