Search papers, labs, and topics across Lattice.
Peking University
1
0
3
VLAs learn to predict task success even when trained only with imitation learning, opening the door to improved performance without expensive reward engineering.
