DRIFT: A Residual Flow Adapter for Decoding Continuous Outputs in Vision-Language Models
返回详情
VLA / Vision-Language-Action 每日论文卡
设备