Towards Brigding the Gap Between Large-scale Pretraining and Efficient Finetuning for Humanoid Control
Ag2x2: A Robust Agent-Agnostic Visual Representation Boosts Zero-Shot Learning of Bimanual Robotic Manipulation