SwiftVLA: Unlocking Spatiotemporal Dynamics for Lightweight VLA Models at Minimal Overhead

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