ReDiverse Restores Diversity Lost in Distilled Diffusion Models
A team from UIUC, fal, and JHU diagnosed the diversity collapse of few-step distilled diffusion models and proposed ReDiverse, a minimal-change fix that restores highly diverse outputs in both image and video models.
2026-11-17 ~ 2026-11-17 · 2 related posts
- ReDiverse restores output diversity in few-step distilled diffusion models — anand_bhattad · 2026-11-17
- ReDiverse fixes diversity collapse in distilled diffusion models via activation steering — ReDiverse · 2026-11-17