Paper · arXiv 2312.00210
DREAM: Diffusion Rectification and Estimation-Adaptive Models
Jinxin Zhou, Tianyu Ding, Tianyi Chen, Jiachen Jiang, Ilya Zharkov, Zhihui Zhu, Luming Liang
DREAM, a novel training framework for diffusion models, improves training alignment with sampling, achieving faster convergence and reduced sampling steps in image super-resolution.
Abstract
We present DREAM, a novel training framework representing Diffusion Rectification and Estimation-Adaptive Models, requiring minimal code changes (just three lines) yet significantly enhancing the alignment of training with sampling in diffusion models. DREAM features two components: diffusion rectification, which adjusts training to reflect the sampling process, and estimation adaptation, which balances perception against distortion. When applied to image super-resolution (SR), DREAM adeptly navigates the tradeoff between minimizing distortion and preserving high image quality. Experiments demonstrate DREAM's superiority over standard diffusion-based SR methods, showing a 2 to 3times faster training convergence and a 10 to 20times reduction in necessary sampling steps to achieve comparable or superior results. We hope DREAM will inspire a rethinking of diffusion model training paradigms.