Bibliographic Details
| Title: |
Multistep Distillation of Diffusion Models via Moment Matching |
| Authors: |
Salimans, Tim, Mensink, Thomas, Heek, Jonathan, Hoogeboom, Emiel |
| Publication Year: |
2024 |
| Collection: |
Computer Science |
| Subject Terms: |
Computer Science - Machine Learning, Computer Science - Artificial Intelligence, Computer Science - Computer Vision and Pattern Recognition, Computer Science - Neural and Evolutionary Computing |
| More Details: |
We present a new method for making diffusion models faster to sample. The method distills many-step diffusion models into few-step models by matching conditional expectations of the clean data given noisy data along the sampling trajectory. Our approach extends recently proposed one-step methods to the multi-step case, and provides a new perspective by interpreting these approaches in terms of moment matching. By using up to 8 sampling steps, we obtain distilled models that outperform not only their one-step versions but also their original many-step teacher models, obtaining new state-of-the-art results on the Imagenet dataset. We also show promising results on a large text-to-image model where we achieve fast generation of high resolution images directly in image space, without needing autoencoders or upsamplers. |
| Document Type: |
Working Paper |
| Access URL: |
http://arxiv.org/abs/2406.04103 |
| Accession Number: |
edsarx.2406.04103 |
| Database: |
arXiv |
