Search papers, labs, and topics across Lattice.
3
0
6
1
Ditch the optimization: MoRe achieves real-time 4D scene reconstruction from monocular video using a feedforward transformer that disentangles motion and structure.
By disentangling structure and motion in the latent space, CoWVLA achieves superior visuomotor learning compared to standard world-model and latent-action approaches.
Skip the computationally expensive propagation of all occupied states in real-time TDDFT: a new equivariant graph transformer, OrbEvo, learns to evolve electronic wavefunctions directly.
