{"id":4367,"date":"2024-02-05T05:03:00","date_gmt":"2024-02-04T20:03:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4367"},"modified":"2024-02-05T05:03:00","modified_gmt":"2024-02-04T20:03:00","slug":"weak-to-strong-jailbreaking-on-large-language-models","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4367","title":{"rendered":"Weak-to-Strong Jailbreaking on Large Language Models\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Weak-to-Strong Jailbreaking on Large Language Models&nbsp;<\/strong>[96.5]<br>Red-teaming\u306e\u30ec\u30dd\u30fc\u30c8\u306b\u3088\u308b\u3068\u3001\u5927\u304d\u306a\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u306f\u3001\u6575\u306e\u30d7\u30ed\u30f3\u30d7\u30c8\u3001\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3001\u30c7\u30b3\u30fc\u30c9\u306b\u3088\u3063\u3066\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u3055\u308c\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u3002 \u672c\u7a3f\u3067\u306f,\u3088\u308a\u5c0f\u578b\u3067\u5b89\u5168\u3067\u306a\u3044LDM\u3092\u7528\u3044\u3066\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u3092\u8a98\u5c0e\u3059\u308b,\u5f31\u5f37\u306e\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u653b\u6483\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2401.17256v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2401.17256v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Tue, 30 Jan 2024 18:48:37 GMT)<\/li>\n\n\n\n<li>\u5f31\u304f\uff08\u5c0f\u3055\u3044\uff09\u30e2\u30c7\u30eb\u306e\u6319\u52d5\u3092\u5206\u6790\u3059\u308b\u3053\u3068\u3067\u5f37\u304f\uff08\u5927\u304d\u3044\uff09\u30e2\u30c7\u30eb\u3092jailbreak\u3067\u304d\u308b\u3068\u306e\u5831\u544a\u3002\u4e0b\u8a18\u306e\u3088\u3046\u306b\u901a\u5e38\u306efine tuning\u3067\u3082\u6709\u52b9\u6027\u304c\u6307\u6458\u3055\u308c\u3066\u3044\u308b\u306e\u3067\u3001jailbreak\u306b\u5fdc\u7528\u3067\u304d\u308b\u3068\u3044\u3046\u306e\u3082\u7d0d\u5f97\u611f\u304c\u3042\u308b\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/XuandongZhao\/weak-to-strong\">XuandongZhao\/weak-to-strong: Weak-to-Strong Jailbreaking on Large Language Models (github.com)<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision\u00a0<\/strong>[55.2]<br>\u8d85\u4eba\u7684\u30e2\u30c7\u30eb\u306f\u3001\u4eba\u9593\u304c\u78ba\u5b9f\u306b\u8a55\u4fa1\u3059\u308b\u3053\u3068\u304c\u96e3\u3057\u3044\u8907\u96d1\u306a\u65b9\u6cd5\u3067\u632f\u308b\u821e\u3046\u3002 \u5f31\u3044\u30e2\u30c7\u30eb\u306e\u76e3\u7763\u306f\u3001\u3088\u308a\u5f37\u529b\u306a\u30e2\u30c7\u30eb\u306e\u5b8c\u5168\u306a\u80fd\u529b\u3092\u5f15\u304d\u51fa\u3059\u3053\u3068\u304c\u3067\u304d\u308b\u304b? \u5f31\u3044\u30e2\u30c7\u30eb\u304c\u751f\u6210\u3057\u305f\u30e9\u30d9\u30eb\u306b\u5f37\u3044\u4e8b\u524d\u8a13\u7df4\u3055\u308c\u305f\u30e2\u30c7\u30eb\u3092\u5fae\u8abf\u6574\u3059\u308b\u3068\u3001\u5f31\u3044\u30b9\u30fc\u30d1\u30fc\u30d0\u30a4\u30b6\u30fc\u3088\u308a\u3082\u4e00\u8cab\u3057\u3066\u6027\u80fd\u304c\u5411\u4e0a\u3059\u308b\u3053\u3068\u304c\u308f\u304b\u3063\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2312.09390v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2312.09390v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 14 Dec 2023 23:07:33 GMT)<\/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":[207,443],"class_list":["post-4367","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-jailbreak","tag-weak-to-strong"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4367","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=4367"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4367\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4367"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4367"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4367"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}