Search papers, labs, and topics across Lattice.
1
0
3
4
Surprisingly, general-purpose vision models already contain better action representations for robotic control than specialized embodied models trained explicitly for that purpose.
