SEEDS: Exponential SDE Solvers for Fast High-Quality Sampling from Diffusion Models

Abstract

A potent class of generative models known as Diffusion Probabilistic Models (DPMs) has become prominent. A forward diffusion process adds gradually noise to data, while a model learns to gradually denoise. Sampling from pre-trained DPMs is obtained by solving differential equations (DE) defined by the learnt model, a process which has shown to be prohibitively slow. Numerous efforts on speeding-up this process have consisted on crafting powerful ODE solvers. Despite being quick, such solvers do not usually reach the optimal quality achieved by available slow SDE solvers. Our goal is to propose SDE solvers that reach optimal quality without requiring several hundreds or thousands of NFEs to achieve that goal. We propose Stochastic Explicit Exponential Derivative-free Solvers (SEEDS), improving and generalizing Exponential Integrator approaches to the stochastic case on several frameworks. After carefully analyzing the formulation of exact solutions of diffusion SDEs, we craft SEEDS to analytically compute the linear part of such solutions. Inspired by the Exponential Time-Differencing method, SEEDS use a novel treatment of the stochastic components of solutions, enabling the analytical computation of their variance, and contains high-order terms allowing to reach optimal quality sampling $\sim 3$-$5\times$ faster than previous SDE methods. We validate our approach on several image generation benchmarks, showing that SEEDS outperform or are competitive with previous SDE solvers. Contrary to the latter, SEEDS are derivative and training free, and we fully prove strong convergence guarantees for them.

arXiv comment60 pages. Camera-Ready version for the 37th Conference on Neural Information Processing Systems (NeurIPS 2023)

References in corpus (20)

• Denoising Diffusion Probabilistic Models
• Diffusion Models Beat GANs on Image Synthesis
• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• Score-Based Generative Modeling through Stochastic Differential Equations
• Elucidating the Design Space of Diffusion-Based Generative Models
• SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations
• Progressive Distillation for Fast Sampling of Diffusion Models
• Pseudo Numerical Methods for Diffusion Models on Manifolds
• Diffusion Models for Adversarial Purification
• Analytic-DPM: an Analytic Estimate of the Optimal Reverse Variance in Diffusion Probabilistic Models
• Fast Sampling of Diffusion Models with Exponential Integrator
• Better Diffusion Models Further Improve Adversarial Training
• gDDIM: Generalized denoising diffusion implicit models
• GENIE: Higher-Order Denoising Diffusion Solvers
• Runge-Kutta Lawson schemes for stochastic differential equations
• Fast Sampling of Diffusion Models via Operator Learning
• Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions
• Robust Principles: Architectural Design Principles for Adversarially Robust CNNs
• Maximum Likelihood Training for Score-Based Diffusion ODEs by High-Order Denoising Score Matching
• Restart Sampling for Improving Generative Processes