Search papers, labs, and topics across Lattice.
3
0
7
K-Forcing accelerates token generation by 2.4-3.5x without abandoning the autoregressive backbone, making it a game-changer for high-load deployments.
State-of-the-art generative models struggle to maintain physical consistency and coherent interactions over time, revealing critical gaps in their world modeling capabilities.
Achieve high-fidelity video generation without compromising reasoning by progressively handing off generation from a lightweight, reasoning-aligned generator to a high-capacity pretrained generator in a shared latent space.
