In this paper, we propose an efficient, fast, and versatile distillation method to accelerate the generation of pre-trained diffusion models: Flash Diffusion. The method reaches state-of-the-art performances in terms of FID and CLIP-Score for few steps image generation on the COCO2014 and COCO2017 datasets, while requiring only several GPU hours of training and fewer trainable parameters than existing methods. In addition to its efficiency, the versatility of the method is also exposed across several tasks such as text-to-image, inpainting, face-swapping, super-resolution and using different backbones such as UNet-based denoisers (SD1.5, SDXL) or DiT (Pixart-$\alpha$), as well as adapters. In all cases, the method allowed to reduce drastically the number of sampling steps while maintaining very high-quality image generation. The official implementation is available at https://github.com/gojasper/flash-diffusion.

翻译：本文提出了一种高效、快速且通用的蒸馏方法，用于加速预训练扩散模型的生成过程：Flash Diffusion。该方法在COCO2014和COCO2017数据集上的少步图像生成任务中，在FID和CLIP-Score指标上达到了最先进的性能，同时仅需数小时GPU训练时间，且可训练参数量少于现有方法。除了高效性之外，该方法在文本到图像、图像修复、人脸替换、超分辨率等多种任务中均展现出卓越的通用性，并能适配UNet架构的去噪器（如SD1.5、SDXL）、DiT架构模型（如Pixart-$\alpha$）以及各类适配器。在所有实验场景中，该方法均能在保持极高图像生成质量的前提下，显著降低采样步数。官方实现代码已发布于https://github.com/gojasper/flash-diffusion。