FutureWorlds · 项目原始材料
核对原文摘录 · 4 段，支持 5 项资料
原阅读方向 / 输出什么
A visual decoder maps predicted tokens back to video.一句话认识
We introduce FutureWorlds, a framework that unifies candidate construction, history maintenance, and learning from relative quality.输入什么
conditioned on the same visual observations, action sequence, and optional language instruction.动作是否影响预测
Robotic world modeling predicts how a scene evolves under actions.