Fast Sampling of Diffusion Models with Exponential Integrator

TL;DR

This paper introduces DEIS, a novel exponential integrator-based sampling method for diffusion models that significantly reduces the number of steps needed for high-quality image generation, achieving state-of-the-art results with fewer evaluations.

Contribution

The paper proposes DEIS, a new sampling algorithm based on exponential integrators, which accelerates diffusion model sampling while maintaining high fidelity, applicable to any diffusion model.

Findings

High-quality samples generated in as few as 10 steps.

Achieves state-of-the-art FID scores with limited score evaluations.

Fast generation of 50k CIFAR10 images in about 3 minutes.

Abstract

The past few years have witnessed the great success of Diffusion models~(DMs) in generating high-fidelity samples in generative modeling tasks. A major limitation of the DM is its notoriously slow sampling procedure which normally requires hundreds to thousands of time discretization steps of the learned diffusion process to reach the desired accuracy. Our goal is to develop a fast sampling method for DMs with a much less number of steps while retaining high sample quality. To this end, we systematically analyze the sampling procedure in DMs and identify key factors that affect the sample quality, among which the method of discretization is most crucial. By carefully examining the learned diffusion process, we propose Diffusion Exponential Integrator Sampler~(DEIS). It is based on the Exponential Integrator designed for discretizing ordinary differential equations (ODEs) and leverages a semilinear structure of the learned diffusion process to reduce the discretization error. The proposed method can be applied to any DMs and can generate high-fidelity samples in as few as 10 steps. In our experiments, it takes about 3 minutes on one A6000 GPU to generate $50k$ images from CIFAR10. Moreover, by directly using pre-trained DMs, we achieve the state-of-art sampling performance when the number of score function evaluation~(NFE) is limited, e.g., 4.17 FID with 10 NFEs, 3.37 FID, and 9.74 IS with only 15 NFEs on CIFAR10. Code is available at https://github.com/qsh-zh/deis