NVIDIA paper: Posterior Matching Distillation for few-step diffusion image generation

nvidia · hf · 2026-11-20

A Virginia Tech × NVIDIA paper (arXiv, Nov 19) proposes Posterior Matching Distillation, which trains a stochastic student to match a teacher diffusion model's full posterior via tilted mean constraints, enabling high-quality few-step image generation without auxiliary score networks. Project page: pmd-distillation.github.io.

Original post →

More from Multimodal

Multimodal channel →