We introduce ForcingBooster, a general framework that dynamically allocates denoising steps to individual patches.
Zekai Gu
EXCAI
AI & ML interests
None yet
Recent Activity
authored a paper about 3 hours ago
ForcingBooster: Adaptive Patch-wise Denoising for Fast Autoregressive Video Diffusion authored a paper about 3 hours ago
PURE: Refining Feed-forward 3D Reconstruction with Unlabeled Post-training authored a paper about 3 hours ago
UniRecGen: Unifying Multi-View 3D Reconstruction and Generation