Search papers, labs, and topics across Lattice.
University of Tsukuba
1
0
2
FlexLAM achieves superior performance over fixed-capacity models by dynamically adjusting latent action lengths, revolutionizing how we approach action alignment in video decision-making.
