Restoring Linguistic Grounding in VLA Models via Train-Free Attention Recalibration

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