用一到多知识蒸馏加速扩散模型，提升生成速度。
Accelerating Diffusion Models with One-to-Many Knowledge Distillation
- 一个教师模型蒸馏出多个学生模型，分时学习不同阶段知识。
- 在多个数据集上实现显著加速，且兼容现有加速方法。
- 适合需要快速生成图像的部署场景，如实时应用。
扩散模型在图像生成方面取得了显著进展，但相比以往生成模型，其计算开销大，难以实现实时生成。现有方法通过改进采样技术或步数蒸馏来减少采样步骤，但每一步的计算成本降低仍研究较少。我们观察到扩散模型在不同时间步具有不同的输入和特征分布，提出一到多知识蒸馏（O2MKD），将单一教师扩散模型蒸馏为多个学生模型，每个学生模型学习教师在特定连续时间步上的知识。在CIFAR10、LSUN Church、CelebA-HQ（基于DDPM）以及COCO30K（基于Stable Diffusion）上的实验表明，O2MKD可与已有知识蒸馏和快速采样方法结合，实现显著加速。代码将在Github发布。
原文摘要 · Abstract (English)
Significant advancements in image generation have been made with diffusion models. Nevertheless, when contrasted with previous generative models, diffusion models face substantial computational overhead, leading to failure in real-time generation. Recent approaches have aimed to accelerate diffusion models by reducing the number of sampling steps through improved sampling techniques or step distillation. However, the methods to diminish the computational cost for each timestep remain a relatively unexplored area. Observing the fact that diffusion models exhibit varying input distributions and feature distributions at different timesteps, we introduce one-to-many knowledge distillation (O2MKD), which distills a single teacher diffusion model into multiple student diffusion models, where each student diffusion model is trained to learn the teacher's knowledge for a subset of continuous timesteps. Experiments on CIFAR10, LSUN Church, CelebA-HQ with DDPM and COCO30K with Stable Diffusion show that O2MKD can be applied to previous knowledge distillation and fast sampling methods to achieve significant acceleration. Codes will be released in Github.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。