{"id":6587,"date":"2025-04-16T04:27:00","date_gmt":"2025-04-15T19:27:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6587"},"modified":"2025-04-16T04:27:00","modified_gmt":"2025-04-15T19:27:00","slug":"deepseek-r1-thoughtology-lets-about-llm-reasoning","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6587","title":{"rendered":"DeepSeek-R1 Thoughtology: Let&#8217;s about LLM Reasoning\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>DeepSeek-R1 Thoughtology: Let&#8217;s &lt;think> about LLM Reasoning\u00a0<\/strong>[31.8]<br>\u672c\u7a3f\u3067\u306f,DeepSeek-R1\u306e\u601d\u8003\u9577,\u9577\u671f\u7684\u30fb\u7d1b\u3089\u308f\u3057\u3044\u6587\u8108\u306e\u7ba1\u7406,\u6587\u5316\u7684\u30fb\u5b89\u5168\u6027\u306b\u95a2\u3059\u308b\u5f71\u97ff\u3068\u5236\u5fa1\u6027\u306b\u3064\u3044\u3066\u691c\u8a0e\u3059\u308b\u3002 DeepSeek-R1\u306b\u306f\u3001\u4f59\u5206\u306a\u63a8\u8ad6\u6642\u9593\u306b\u3088\u3063\u3066\u30e2\u30c7\u30eb\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u304c\u640d\u306a\u308f\u308c\u308b\u3088\u3046\u306a\u63a8\u8ad6\u306e\u2018\u30b9\u30a4\u30fc\u30c8\u30b9\u30dd\u30c3\u30c8\u2019\u304c\u3042\u308b\u3002 \u307e\u305f\u3001DeepSeek-R1\u306e\u5b89\u5168\u6027\u4e0a\u306e\u8106\u5f31\u6027\u306f\u3001\u975e\u5408\u7406\u7684\u306a\u8106\u5f31\u6027\u3068\u6bd4\u3079\u3066\u3082\u5927\u304d\u3044\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2504.07128v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2504.07128v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 02 Apr 2025 00:36:08 GMT)<\/li>\n\n\n\n<li>DeepSeek R1\u306e\u63a8\u8ad6\u306b\u95a2\u3059\u308b\u5206\u6790\u3001\u300cDeepSeek-R1 exhibits higher safety vulnerabilities compared to its non-reasoning counterpart DeepSeek-V3 (DeepSeek-AI et al , 2025b).\u3000We also show that the model\u2019s reasoning capabilities can be used to generate jailbreak attacks that successfully elicit harmful responses from safety-aligned LLMs.\u300d\u3001\u300cWhen presented with moral or cultural questions, DeepSeek-R1 reasons for significantly longer when prompted in English than when prompted in Chinese. It also provides different responses, displaying different sets of cultural values in each language\u300d\u306f\u9762\u767d\u3044\u3002<\/li>\n<\/ul>\n\n\n\n<p><\/p>\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":[232,283,366],"class_list":["post-6587","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-lrm","tag-o1","tag-slow-thinking"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6587","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=6587"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6587\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6587"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6587"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6587"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}