{"id":4187,"date":"2023-12-18T04:23:00","date_gmt":"2023-12-17T19:23:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4187"},"modified":"2023-12-18T04:23:00","modified_gmt":"2023-12-17T19:23:00","slug":"cyberseceval","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4187","title":{"rendered":"CyberSecEval"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Models\u00a0<\/strong>[41.1]<br>\u672c\u7a3f\u3067\u306f,Large Language Models (LLMs) \u306e\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u30a2\u30b7\u30b9\u30bf\u30f3\u30c8\u3068\u3057\u3066\u306e\u30b5\u30a4\u30d0\u30fc\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u3092\u4fc3\u9032\u3059\u308b\u305f\u3081\u306b\u958b\u767a\u3055\u308c\u305f,\u5305\u62ec\u7684\u306a\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u3042\u308bCyberSecEval\u3092\u63d0\u6848\u3059\u308b\u3002 CyberSecEval\u306f\u30012\u3064\u306e\u91cd\u8981\u306a\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u9818\u57df\u306b\u304a\u3051\u308bLSM\u306e\u5fb9\u5e95\u7684\u306a\u8a55\u4fa1\u3092\u63d0\u4f9b\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2312.04724v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2312.04724v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 7 Dec 2023 22:07:54 GMT)<\/li>\n\n\n\n<li>\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u95a2\u9023\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3068\u3057\u3066\u300c\u5b89\u5168\u3067\u306a\u3044\u30b3\u30fc\u30c9\u306e\u751f\u6210\u300d\u300c\u30b5\u30a4\u30d0\u30fc\u653b\u6483\u306e\u652f\u63f4\u306b\u5bfe\u3059\u308b\u30b3\u30f3\u30d7\u30e9\u30a4\u30a2\u30f3\u30b9\u300d\u3092\u8a55\u4fa1\u3059\u308b\u3082\u306e\u3002<a href=\"https:\/\/ai.meta.com\/research\/publications\/purple-llama-cyberseceval-a-benchmark-for-evaluating-the-cybersecurity-risks-of-large-language-models\/\">Purple Llama CyberSecEval: A benchmark for evaluating the cybersecurity risks of large language models | Research &#8211; AI at Meta<\/a>\u306e\u7acb\u3061\u4e0a\u3052\u306b\u4f34\u3046\u3082\u306e\u3002<\/li>\n\n\n\n<li>\u300cOn average, LLMs suggested vulnerable code 30% of the time over CYBERSECEVAL \u2019s test cases. Furthermore, models complied with 53% of requests to assist in cyberattacks on average across all models and threat categories.\u300d\u3068\u306e\u3053\u3068\u3067\u9053\u306f\u306a\u304c\u305d\u3046\u3002GPT-4\u3067\u3042\u308c\u3070\u5927\u4e08\u592b\u3068\u3044\u3046\u7d50\u679c\u3067\u3082\u306a\u3044\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/facebookresearch\/PurpleLlama\/tree\/main\/CybersecurityBenchmarks\">PurpleLlama\/CybersecurityBenchmarks at main \u00b7 facebookresearch\/PurpleLlama \u00b7 GitHub<\/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":[223,483,517,622],"class_list":["post-4187","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-483","tag-517","tag-622"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4187","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=4187"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4187\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4187"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4187"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4187"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}