{"id":32927,"date":"2025-04-12T14:16:41","date_gmt":"2025-04-12T06:16:41","guid":{"rendered":"https:\/\/www.1ai.net\/?p=32927"},"modified":"2025-04-12T14:16:41","modified_gmt":"2025-04-12T06:16:41","slug":"%e8%8b%b1%e4%bc%9f%e8%be%be%e5%8f%91%e5%b8%83-llama-3-1-nemotron-ultra-253b-v1-%e6%a8%a1%e5%9e%8b%ef%bc%8c%e6%8e%a8%e5%8a%a8-ai-%e9%ab%98%e6%95%88%e9%83%a8%e7%bd%b2","status":"publish","type":"post","link":"http:\/\/www.1ai.net\/en\/32927.html","title":{"rendered":"NVIDIA Releases Llama-3.1-Nemotron-Ultra-253B-v1 Model to Drive Efficient AI Deployment"},"content":{"rendered":"<p>4 \u6708 12 \u65e5\u6d88\u606f\uff0c\u79d1\u6280\u5a92\u4f53 marktechpost \u6628\u65e5\uff084 \u6708 11 \u65e5\uff09\u53d1\u5e03\u535a\u6587\uff0c\u62a5\u9053\u79f0<a href=\"http:\/\/www.1ai.net\/en\/tag\/%e8%8b%b1%e4%bc%9f%e8%be%be\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u82f1\u4f1f\u8fbe]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u82f1\u4f1f\u8fbe<\/a>\u53d1\u5e03\u00a0Llama-3.1-Nemotron-Ultra-253B-v1\uff0c<strong>\u8fd9\u6b3e 2530 \u4ebf\u53c2\u6570\u7684<a href=\"http:\/\/www.1ai.net\/en\/tag\/%e5%a4%a7%e5%9e%8b%e8%af%ad%e8%a8%80%e6%a8%a1%e5%9e%8b\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u5927\u578b\u8bed\u8a00\u6a21\u578b]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u5927\u578b\u8bed\u8a00\u6a21\u578b<\/a>\u5728\u63a8\u7406\u80fd\u529b\u3001\u67b6\u6784\u6548\u7387\u548c\u751f\u4ea7\u51c6\u5907\u5ea6\u4e0a\u5b9e\u73b0\u91cd\u5927\u7a81\u7834\u3002<\/strong><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-32928\" title=\"1c37f365j00sulddy004jd000v900h4p\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/04\/1c37f365j00sulddy004jd000v900h4p.jpg\" alt=\"1c37f365j00sulddy004jd000v900h4p\" width=\"1125\" height=\"616\" \/><\/p>\n<p>\u968f\u7740 AI \u5728\u6570\u5b57\u57fa\u7840\u8bbe\u65bd\u4e2d\u7684\u666e\u53ca\uff0c\u4f01\u4e1a\u4e0e\u5f00\u53d1\u8005\u9700\u5728\u8ba1\u7b97\u6210\u672c\u3001\u6027\u80fd\u4e0e\u6269\u5c55\u6027\u95f4\u5bfb\u627e\u5e73\u8861\u3002\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u7684\u5feb\u901f\u53d1\u5c55\u63d0\u5347\u4e86\u81ea\u7136\u8bed\u8a00\u7406\u89e3\u548c\u5bf9\u8bdd\u80fd\u529b\uff0c\u4f46\u5176\u5e9e\u5927\u89c4\u6a21\u5e38\u5bfc\u81f4\u6548\u7387\u4f4e\u4e0b\uff0c\u9650\u5236\u5927\u89c4\u6a21\u90e8\u7f72\u3002<\/p>\n<p>\u82f1\u4f1f\u8fbe\u6700\u65b0\u53d1\u5e03\u7684\u00a0Llama-3.1-Nemotron-Ultra-253B-v1\uff08\u7b80\u79f0 Nemotron Ultra\uff09\u76f4\u9762\u8fd9\u4e00\u6311\u6218\uff0c\u8be5\u6a21\u578b\u57fa\u4e8e Meta \u7684 Llama-3.1-405B-Instruct \u67b6\u6784\uff0c\u4e13\u4e3a\u5546\u4e1a\u548c\u4f01\u4e1a\u9700\u6c42\u8bbe\u8ba1\uff0c\u652f\u6301\u4ece\u5de5\u5177\u4f7f\u7528\u5230\u591a\u8f6e\u590d\u6742\u6307\u4ee4\u6267\u884c\u7b49\u4efb\u52a1\u3002<\/p>\n<p>1AI\u63f4\u5f15\u535a\u6587\u4ecb\u7ecd\uff0cNemotron Ultra \u91c7\u7528\u4ec5\u89e3\u7801\u5668\u7684\u5bc6\u96c6 Transformer \u7ed3\u6784\uff0c\u901a\u8fc7\u795e\u7ecf\u67b6\u6784\u641c\u7d22\uff08NAS\uff09\u7b97\u6cd5\u4f18\u5316\uff0c\u5176\u521b\u65b0\u4e4b\u5904\u5728\u4e8e\u91c7\u7528\u8df3\u8dc3\u6ce8\u610f\u529b\u673a\u5236\uff0c\u5728\u90e8\u5206\u5c42\u4e2d\u7701\u7565\u6ce8\u610f\u529b\u6a21\u5757\u6216\u66ff\u6362\u4e3a\u7b80\u5355\u7ebf\u6027\u5c42\u3002<\/p>\n<p>\u6b64\u5916\uff0c\u524d\u9988\u7f51\u7edc\uff08FFN\uff09\u878d\u5408\u6280\u672f\u5c06\u591a\u5c42 FFN \u5408\u5e76\u4e3a\u66f4\u5bbd\u4f46\u66f4\u5c11\u7684\u5c42\uff0c\u5927\u5e45\u7f29\u77ed\u63a8\u7406\u65f6\u95f4\uff0c\u540c\u65f6\u4fdd\u6301\u6027\u80fd\u3002\u6a21\u578b\u652f\u6301 128K token \u7684\u4e0a\u4e0b\u6587\u7a97\u53e3\uff0c\u53ef\u5904\u7406\u957f\u7bc7\u6587\u672c\uff0c\u9002\u5408\u9ad8\u7ea7 RAG \u7cfb\u7edf\u548c\u591a\u6587\u6863\u5206\u6790\u3002<\/p>\n<p>\u5728\u90e8\u7f72\u6548\u7387\u4e0a\uff0cNemotron Ultra \u4e5f\u5b9e\u73b0\u7a81\u7834\u3002\u5b83\u80fd\u5728\u5355 8xH100 \u8282\u70b9\u4e0a\u8fd0\u884c\u63a8\u7406\uff0c\u663e\u8457\u964d\u4f4e\u6570\u636e\u4e2d\u5fc3\u6210\u672c\uff0c\u63d0\u5347\u4f01\u4e1a\u5f00\u53d1\u8005\u7684\u53ef\u53ca\u6027\u3002<\/p>\n<p>\u82f1\u4f1f\u8fbe\u901a\u8fc7\u591a\u9636\u6bb5\u540e\u8bad\u7ec3\u8fdb\u4e00\u6b65\u4f18\u5316\u6a21\u578b\uff0c\u5305\u62ec\u5728\u4ee3\u7801\u751f\u6210\u3001\u6570\u5b66\u3001\u5bf9\u8bdd\u548c\u5de5\u5177\u8c03\u7528\u7b49\u4efb\u52a1\u4e0a\u7684\u76d1\u7763\u5fae\u8c03\uff0c\u4ee5\u53ca\u4f7f\u7528\u7fa4\u4f53\u76f8\u5bf9\u7b56\u7565\u4f18\u5316\uff08GRPO\uff09\u7b97\u6cd5\u8fdb\u884c\u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\u3002\u8fd9\u4e9b\u6b65\u9aa4\u786e\u4fdd\u6a21\u578b\u5728\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8868\u73b0\u51fa\u8272\uff0c\u5e76\u4e0e\u4eba\u7c7b\u4ea4\u4e92\u504f\u597d\u9ad8\u5ea6\u5951\u5408\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>April 12, technology media marktechpost yesterday (April 11) published a blog post, reported that NVIDIA released Llama-3.1-Nemotron-Ultra-253B-v1, a 253 billion parameter large-scale language model that achieves a major breakthrough in inference capability, architectural efficiency and production readiness. As AI becomes pervasive in digital infrastructure, organizations and developers need to find a balance between computational cost, performance, and scalability. The rapid development of large-scale language models (LLMs) has improved natural language understanding and dialog capabilities, but their large scale often leads to inefficiencies that limit large-scale deployments. NVIDIA's newly released Llama-3.1-Nemotron-Ultr<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[146],"tags":[371,239],"collection":[],"class_list":["post-32927","post","type-post","status-publish","format-standard","hentry","category-news","tag-371","tag-239"],"acf":[],"_links":{"self":[{"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/32927","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=32927"}],"version-history":[{"count":0,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/32927\/revisions"}],"wp:attachment":[{"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=32927"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=32927"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=32927"},{"taxonomy":"collection","embeddable":true,"href":"http:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=32927"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}