Search papers, labs, and topics across Lattice.
1
0
2
5
Off-policy RL can now train in minutes instead of hours for sim-to-real robot control, thanks to a new algorithm that borrows scaling laws from supervised learning.
