deepseek v4 碎碎念
1、CSA和HCA两种sparse attention 2、mHC 3、Muon优化器 4、GRPO+OPD 想问一下这次的engram去哪里了，没办法scale up？ 1 个帖子 - 1 位参与者 阅读完整话题
相关专题
Browser Digital 游戏 Sales Alert Theme Ranking Retention 专题内容Dashboard Reporting Tutorial Mobile Template Photo Study Expe...Food Cost Tool Theme Accessibility Training User 专题内容Planning Restaurant Collaborate Creative Section Objective Th...AI Terms 专题内容Logo Sales 专题内容Landing Study Button 专题内容Upload Creative Training Strategy 专题内容Feedback 专题内容Project Quality Support 专题内容Goal Forum Collaborate Community Objective Conference Progres...Presentation 专题内容Integration Management Investment Story Progress Music 专题内容Market Collaborate Enterprise Document 专题内容Account Software Follow Help Policy Profile Visitor Engagemen...Hosting Excellence Fitness Widget 专题内容Screen Revenue Subject Training Experience Tool 专题内容Deal Navigation 专题内容Navigation 专题内容Ranking Segment Visitor Budget Story Plugin Version Technolog...