CompVis improves Distributional Diffusion Models: 4.48 FID at 4 steps on ImageNet
CompVis · hf · 2026-09-30
CompVis releases Improved Distributional Diffusion Models (iDDM). DDMs replace mean-prediction denoisers with distributional denoisers trained via a scoring rule, but faced two scaling obstacles: multi-particle training overhead and globally fixed scoring-rule hyperparameters forcing one trade-off across sampling budgets.
The improvements defer particle expansion to late transformer layers and introduce time-dependent scoring rule schedules informed by the dynamical regimes of Biroli 2024. With a DiT-based latent setup, a DiT-XL/2 trained from scratch in one stage (no teacher, self-distillation or JVPs) reaches 4.48 FID at 4 steps and 2.38 at 50 steps on ImageNet-256², with FID not degrading as budget grows from 4 to 50 NFE. The recipe transfers to text-to-image. Code and models are open-sourced on GitHub.
More from Multimodal
- Runway shows how to turn one product photo into a full social ad campaign — tlakomy · 2026-09-30
- Pencil0 enters private beta: an AI filmmaking agent that turns one logline into a full film — yihui_indie · 2026-09-30
- SoL-Refiner turns low-res AI video into 2K/4K in one denoising step, 8.91x faster — linoy_tsaban · 2026-09-30
- Weekend AI experiment grows ever-changing haikus from words you water — AI_Andrew · 2026-09-30
- From deer videos to a 4D time machine: the 'spatial memory palace' for physical reality — bilawalsidhu · 2026-09-30
- New AA-Video-T2V v2.0 Benchmark Ranks Wan 3.0 #1, MiniMax H3 at 1/7 the Price of Seedance — ArtificialAnlys · 2026-09-30