{"id":3384,"date":"2023-05-29T05:46:00","date_gmt":"2023-05-28T20:46:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3384"},"modified":"2023-05-29T05:46:00","modified_gmt":"2023-05-28T20:46:00","slug":"rwkv-reinventing-rnns-for-the-transformer-era","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3384","title":{"rendered":"RWKV: Reinventing RNNs for the Transformer Era"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>RWKV: Reinventing RNNs for the Transformer Era\u00a0<\/strong>[27.3]<br>\u672c\u7a3f\u3067\u306f,\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u306e\u52b9\u7387\u7684\u306a\u4e26\u5217\u5316\u5b66\u7fd2\u3068RNN\u306e\u52b9\u7387\u7684\u306a\u63a8\u8ad6\u3092\u7d44\u307f\u5408\u308f\u305b\u305f\u65b0\u3057\u3044\u30e2\u30c7\u30eb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u63d0\u6848\u3059\u308b\u3002 \u63d0\u6848\u624b\u6cd5\u306f\u7dda\u5f62\u30a2\u30c6\u30f3\u30b7\u30e7\u30f3\u6a5f\u69cb\u3092\u5229\u7528\u3057\u3066,\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u4e2d\u306b\u8a08\u7b97\u3092\u4e26\u5217\u5316\u3057,\u63a8\u8ad6\u4e2d\u306b\u4e00\u5b9a\u306e\u8a08\u7b97\u91cf\u3068\u30e1\u30e2\u30ea\u306e\u8907\u96d1\u3055\u3092\u7dad\u6301\u3059\u308b\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u3042\u308b\u3044\u306fRNN\u3068\u3057\u3066\u30e2\u30c7\u30eb\u3092\u5b9a\u5f0f\u5316\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u3002 \u6211\u3005\u306e\u5b9f\u9a13\u306f\u3001RWKV\u304c\u540c\u69d8\u306e\u5927\u304d\u3055\u306e\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u3068\u540c\u7b49\u306b\u52d5\u4f5c\u3057\u3066\u3044\u308b\u3053\u3068\u3092\u793a\u3057\u3001\u5c06\u6765\u306e\u4f5c\u696d\u304c\u3053\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u6d3b\u7528\u3057\u3066\u3088\u308a\u52b9\u7387\u7684\u306a\u30e2\u30c7\u30eb\u3092\u4f5c\u6210\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u3053\u3068\u3092\u793a\u5506\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2305.13048v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2305.13048v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 22 May 2023 13:57:41 GMT)<\/li>\n\n\n\n<li>\u6027\u80fd\u304c\u9ad8\u3044\u3068\u5642\u306eRNN\u30d9\u30fc\u30b9\u306eRWKV\u306e\u8ad6\u6587<\/li>\n\n\n\n<li>\u300cWhile many alternatives to Transformers have been proposed with similar claims, ours is the first to back up those claims with pretrained models with tens of billions of parameters.\u300d\u3068\u3044\u3046\u8a18\u8f09\u304c\u71b1\u304f\u3001\u304a\u3063\u3057\u3083\u308b\u901a\u308a\u3067\u5b9f\u7528\u30ec\u30d9\u30eb\u306e\u5927\u304d\u3055\uff06\u6709\u540d\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u6709\u52b9\u6027\u3092\u793a\u3059\u3053\u3068\u306f\u91cd\u8981\u3060\u3068\u601d\u3046\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/BlinkDL\/RWKV-LM\">GitHub &#8211; BlinkDL\/RWKV-LM: RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it&#8217;s combining the best of RNN and transformer &#8211; great performance, fast inference, saves VRAM, fast training, &#8220;infinite&#8221; ctx_len, and free sentence embedding.<\/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":[223,341],"class_list":["post-3384","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-rnn"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3384","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=3384"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3384\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3384"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3384"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3384"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}