World2Act: Latent Action Post-Training via Skill-Compositional World Models

返回详情 VLA / Vision-Language-Action 每日论文卡