Search papers, labs, and topics across Lattice.
Affiliation:
3
0
5
A scalar adjoint method slashes the computational cost of fine-tuning flow policies in RL, achieving state-of-the-art results on complex manipulation tasks.
Swapping objects in 3D space can boost VLA policy success rates by over 16% on novel objects without additional data collection.
Learning dexterous manipulation policies from human videos can achieve a 75% success rate on real-robot tasks without any robot-specific demonstrations.
