Researcher, Stanford University
2 papers at NeurIPS 2025
a sparse attention with $\mathcal O(n \log n)$ complexity for long video generation
We propose a method to speedup video diffusion generation through efficient attention.