量子位·2026-09-23 15:29·国内权威源DeepSeek新论文公开Agent训练!梁文锋署名中文摘要DeepSeek发布由梁文锋署名的Agent训练新论文,每秒可生成5000多个沙盒。English SummaryDeepSeek's new Agent training paper, signed by Liang Wenfeng, can generate over 5,000 sandboxes per second.原文节选每秒能产生5000+个沙盒阅读原文View in English