Regularized Distribution Matching Distillation for One-Step Unpaired Image-to-Image Translation
Abstract
Diffusion distillation methods aim to compress the diffusion models into efficient one-step generators while trying to preserve quality. Among them, Distribution Matching Distillation (DMD) offers a suitable framework for training general-form one-step generators, applicable beyond unconditional generation. In this work, we introduce its modification, called Regularized Distribution Matching Distillation, applicable to unpaired image-to-image problems. We demonstrate its empirical performance in application to several translation tasks, including 2D examples and I2I between different image datasets, where it performs on par or better than multi-step diffusion baselines.
Cite
Text
Rakitin et al. "Regularized Distribution Matching Distillation for One-Step Unpaired Image-to-Image Translation." ICML 2024 Workshops: SPIGM, 2024.Markdown
[Rakitin et al. "Regularized Distribution Matching Distillation for One-Step Unpaired Image-to-Image Translation." ICML 2024 Workshops: SPIGM, 2024.](https://mlanthology.org/icmlw/2024/rakitin2024icmlw-regularized/)BibTeX
@inproceedings{rakitin2024icmlw-regularized,
title = {{Regularized Distribution Matching Distillation for One-Step Unpaired Image-to-Image Translation}},
author = {Rakitin, Denis and Shchekotov, Ivan and Vetrov, Dmitry},
booktitle = {ICML 2024 Workshops: SPIGM},
year = {2024},
url = {https://mlanthology.org/icmlw/2024/rakitin2024icmlw-regularized/}
}