Search papers, labs, and topics across Lattice.
1
0
2
Self-supervised finetuning can preserve a VLA model's instruction-following abilities while significantly boosting its performance on new robotic tasks.
