{"id":4776,"date":"2024-04-26T05:43:00","date_gmt":"2024-04-25T20:43:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4776"},"modified":"2024-04-26T05:43:00","modified_gmt":"2024-04-25T20:43:00","slug":"physics-of-language-models-part-3-3-knowledge-capacity-scaling-laws","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4776","title":{"rendered":"Physics of Language Models: Part 3.3, Knowledge Capacity Scaling Laws\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Physics of Language Models: Part 3.3, Knowledge Capacity Scaling Laws\u00a0<\/strong>[51.7]<br>\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u6cd5\u5247\u306f\u3001\u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u30b5\u30a4\u30ba\u3068\u80fd\u529b\u306e\u95a2\u4fc2\u3092\u8a18\u8ff0\u3057\u3066\u3044\u308b\u3002 \u6211\u3005\u306f\u3001\u30a6\u30a3\u30ad\u30da\u30c7\u30a3\u30a2\u306e\u30da\u30fc\u30b8\u304b\u3089(\u7c73\u56fd\u3001\u9996\u90fd\u30ef\u30b7\u30f3\u30c8\u30f3D.C.\u306a\u3069)\u30c9\u30e1\u30a4\u30f3\u3068\u3057\u3066\u8868\u3055\u308c\u308b\u4e8b\u5b9f\u77e5\u8b58\u306b\u7126\u70b9\u3092\u5f53\u3066\u308b\u3002 7B\u30e2\u30c7\u30eb\u306f\u3001\u82f1\u8a9e\u306e\u30a6\u30a3\u30ad\u30da\u30c7\u30a3\u30a2\u3068\u6559\u79d1\u66f8\u3092\u5408\u308f\u305b\u305f14B\u30d3\u30c3\u30c8\u306e\u77e5\u8b58\u3092\u4fdd\u5b58\u3067\u304d\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2404.05405v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2404.05405v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 08 Apr 2024 11:11:31 GMT)<\/li>\n\n\n\n<li>\u300cThrough multiple controlled datasets, we establish that language models can and only can store 2 bits of knowledge per parameter, even when quantized to int8, and such knowledge can be flexibly extracted for downstream applications.\u300d\u3068\u306e\u3053\u3068\u3002\u9762\u767d\u3044\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":[223],"class_list":["post-4776","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4776","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=4776"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4776\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4776"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4776"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4776"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}