Search papers, labs, and topics across Lattice.
1
0
2
13
A unified Vision-Language Model and Diffusion architecture unlocks surprisingly effective optical flow forecasting from noisy web data, enabling language-conditioned robot control and video generation.
