{"id":1725,"date":"2022-04-07T05:00:00","date_gmt":"2022-04-06T20:00:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=1725"},"modified":"2022-04-07T05:00:00","modified_gmt":"2022-04-06T20:00:00","slug":"palm-pathways-language-model","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=1725","title":{"rendered":"PaLM: Pathways Language Model"},"content":{"rendered":"\n<ul class=\"wp-block-list\"><li><strong>PaLM: Scaling Language Modeling with Pathways\u00a0<\/strong>[180.7]<br>\u6211\u3005\u306f,\u30d1\u30b9\u30a6\u30a7\u30a4\u30ba\u8a00\u8a9e\u30e2\u30c7\u30eb PaLM \u3068\u547c\u3070\u308c\u308b\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u8a00\u8a9e\u30e2\u30c7\u30eb\u3092\u7528\u3044\u3066,540\u30d3\u30ea\u30aa\u30f3\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u8a13\u7df4\u3057\u305f\u3002 \u6211\u3005\u306fPathways\u3068\u3044\u3046\u65b0\u3057\u3044ML\u30b7\u30b9\u30c6\u30e0\u3092\u7528\u3044\u3066,6144 TPU v4\u30c1\u30c3\u30d7\u4e0a\u3067PaLM\u3092\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u3002 \u6570\u767e\u306e\u8a00\u8a9e\u7406\u89e3\u304a\u3088\u3073\u751f\u6210\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u4e0a\u3067\u3001\u6700\u5148\u7aef\u306e\u6570\u767a\u306e\u5b66\u7fd2\u7d50\u679c\u3092\u9054\u6210\u3057\u3001\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u306e\u7d99\u7d9a\u7684\u306a\u30e1\u30ea\u30c3\u30c8\u3092\u5b9f\u8a3c\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2204.02311v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2204.02311v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 5 Apr 2022 16:11:45 GMT)<ul><li>540-billion parameter\u3067780 billion tokens\u00a0\u306e\u30c7\u30fc\u30bf\uff06\u00a06144\u500b\u306eTPU v4 \u3092\u7528\u3044\u3066\u69cb\u7bc9\u3055\u308c\u305f\u5927\u898f\u6a21\u30e2\u30c7\u30eb\u3002BIG-bench\u3067\u5e73\u5747\u7684\u306a\u4eba\u9593\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u4e0a\u56de\u308b\u3002<\/li><li>Discontinuous improvements\u3068\u3057\u3066\u5831\u544a\u3055\u308c\u305f\u5185\u5bb9\u304c\u8208\u5473\u6df1\u304f\u30018B\u219262B\u30d1\u30e9\u30e1\u30fc\u30bf\u3078\u306e\u6539\u5584\u306862B\u2192540B\u3078\u306e\u6539\u5584\u306b\u304a\u3044\u3066\u3088\u304f\u5831\u544a\u3055\u308c\u308b\u201cpower law\u201d\u306b\u6cbf\u3063\u305f\u6539\u5584\u3067\u306f\u306a\u3044\u3001\u975e\u9023\u7d9a\u7684\u306a\u6539\u5584\u304c\u898b\u3089\u308c\u305f\u3068\u306e\u3053\u3068\u3002<\/li><li>\u300cFirst, the results presented here suggest that the improvements from scale for few-shot language understanding have not yet plateaued.\u300d\u3068\u3042\u308b\u901a\u308a\u307e\u3060\u767a\u5c55\u304c\u898b\u8fbc\u3081\u308b\u3068\u306e\u3053\u3068\u3067\u9762\u767d\u3044\u3002<\/li><\/ul><\/li><\/ul>\n","protected":false},"excerpt":{"rendered":"<p>PaLM: Scaling Language Modeling with Pathways\u00a0[180.7]\u6211\u3005\u306f,\u30d1\u30b9\u30a6\u30a7\u30a4\u30ba\u8a00\u8a9e\u30e2\u30c7\u30eb PaLM \u3068\u547c\u3070\u308c\u308b\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u8a00\u8a9e\u30e2\u30c7\u30eb\u3092\u7528\u3044\u3066,540\u30d3\u30ea\u30aa\u30f3\u306e\u30d1\u30e9\u30e1\u30fc &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=1725\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;PaLM: Pathways Language Model&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[146,170,567],"class_list":["post-1725","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-fewshot","tag-gigantic-language-model","tag-567"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/1725","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=1725"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/1725\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1725"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1725"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1725"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}