{"url":"https://yuzukaki-chemical.hateblo.jp/entry/2025/06/21/164613","provider_url":"https://hatena.blog","version":"1.0","published":"2025-06-21 16:46:13","blog_url":"https://yuzukaki-chemical.hateblo.jp/","width":"100%","blog_title":"\u3086\u305a\u304b\u304d\u306e\u30d6\u30ed\u30b0","image_url":null,"author_name":"yuzukaki1000","categories":["\u30d7\u30ed\u30f3\u30d7\u30c8\u30a8\u30f3\u30b8\u30cb\u30a2\u30ea\u30f3\u30b0"],"html":"<iframe src=\"https://hatenablog-parts.com/embed?url=https%3A%2F%2Fyuzukaki-chemical.hateblo.jp%2Fentry%2F2025%2F06%2F21%2F164613\" title=\"\u3010\u8ad6\u6587\u8aad\u89e3\u3011DeepSeek R1 \u5fb9\u5e95\u89e3\u8aac\uff1a\u5f37\u5316\u5b66\u7fd2\u306e\u307f\u3067LLM\u306e\u591a\u6bb5\u63a8\u8ad6\u304c\u3053\u3053\u307e\u3067\u9032\u5316\uff01\u8a71\u984c\u306eAI\u30e2\u30c7\u30eb\u3092\u8ad6\u6587\u89e3\u8aac - \u3086\u305a\u304b\u304d\u306e\u30d6\u30ed\u30b0\" class=\"embed-card embed-blogcard\" scrolling=\"no\" frameborder=\"0\" style=\"display: block; width: 100%; height: 190px; max-width: 500px; margin: 10px 0px;\"></iframe>","type":"rich","author_url":"https://blog.hatena.ne.jp/yuzukaki1000/","title":"\u3010\u8ad6\u6587\u8aad\u89e3\u3011DeepSeek R1 \u5fb9\u5e95\u89e3\u8aac\uff1a\u5f37\u5316\u5b66\u7fd2\u306e\u307f\u3067LLM\u306e\u591a\u6bb5\u63a8\u8ad6\u304c\u3053\u3053\u307e\u3067\u9032\u5316\uff01\u8a71\u984c\u306eAI\u30e2\u30c7\u30eb\u3092\u8ad6\u6587\u89e3\u8aac","height":"190","description":"\u3053\u3093\u306b\u3061\u306f\u3001\u3086\u305a\u304b\u304d\u3067\u3059\u3002 \u4eca\u56de\u306f\u3001\u300cDeepSeek R1\u300d \u3068\u547c\u3070\u308c\u308b\u3001\u6700\u65b0\u306e\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM\uff09\u306e\u63a8\u8ad6\u80fd\u529b\u3092\u5f37\u5316\u3059\u308b\u305f\u3081\u306e\u7814\u7a76\u306b\u3064\u3044\u3066\u3001\u4e00\u3064\u306e\u6280\u8853\u30d6\u30ed\u30b0\u8a18\u4e8b\u3068\u3057\u3066\u307e\u3068\u3081\u3066\u307f\u305f\u3044\u3068\u601d\u3044\u307e\u3059\u3002 \u4eca\u56de\u53d6\u308a\u4e0a\u3052\u308b\u300cDeepSeek R1\u300d\u306f\u3001arXiv\u306b\u3066\u516c\u958b\u3055\u308c\u305f\u8ad6\u6587 DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning (arXiv:2501.12948) \u304b\u3089\u306e\u5185\u5bb9\u3092\u30d9\u30fc\u30b9\u3068\u3057\u3066\u3044\u307e\u3059\u3002\u3068\u304f\u306b\u300cDeepSeek-R1-Zero\u300d\u3068\u3044\u3046\u30e2\u30c7\u30eb\u304c\u3001\u4e8b\u524d\u306e\u6559\u5e2b\u3042\u308a\u5b66\u7fd2\uff08SFT\uff09\u3092\u2026","provider_name":"Hatena Blog"}