Search papers, labs, and topics across Lattice.
Affiliation:
4
0
10
A single latent cache can unify diverse visual representations and spatiotemporal queries for efficient, consistent dynamic world modeling across multiple video generation tasks.
MLLMs excel at reasoning over diagrams but falter in parsing and editing them, revealing a significant gap that needs addressing.
Zero-initialization emerges as the key driver behind the superior performance of adaLN-Zero in diffusion transformers, reshaping our understanding of conditioning mechanisms in image generation.
