Saining Xie· @sainingxie · X·· 2026-03-05
AI 导读
来自 @TongPetersb、@DavidJFan 和 @__JohnNguyen__ 的又一项科学探索，即使你身处前沿实验室，也可能会让你学到新东西 这里有很多有趣的观察，但我只强调一点： - 尝试用 MoE 扩展 DiTs 大多徒劳无功，这算是行业公开的秘密。 - 但 RAE 与 MoE 之间意外却直观的协同作用，可能真的会改变这一点。 [引用 @TongPetersb]：超越语言训练。我们押注视觉世界，将其作为与语言建模并行且超越它的关键下一步。因此，我们研究了从零开始用视觉构建基础模型。我们分享我们的探索：视觉表征、数据、世界建模、架构和扩展行为！[1/9]
正文 · 原文
another scientific exploration from @TongPetersb, @DavidJFan, and @__JohnNguyen__ that might teach you something new, even if you’re in a frontier lab
lots of interesting observations here, but I’ll highlight just one:
- it’s kind of an open industry secret that trying to scale DiTs with MoE has mostly been fruitless.
- the unexpected, yet intuitive, synergy between RAE and MoE might actually change that.
Train Beyond Language. We bet on the visual world as the critical next step alongside and beyond language modeling. So, we studied building foundation models from scratch with vision. We share our exploration: visual representations, data, world modeling, architecture, and scaling behavior! [1/9]在 X 查看被引用的帖子
来源：Saining Xie · x.com