Efficient Integrators for Diffusion Generative Models
This work addresses the inference-time efficiency bottleneck for diffusion models, offering incremental improvements in sampling speed and quality.
The paper tackles the slow sample generation problem in diffusion models by proposing two complementary frameworks, Conjugate Integrators and Splitting Integrators, which accelerate sampling in pre-trained models. The result is improved performance, with deterministic and stochastic samplers achieving FID scores of 2.11 and 2.36 in 100 NFE on CIFAR-10, outperforming baselines.
Diffusion models suffer from slow sample generation at inference time. Therefore, developing a principled framework for fast deterministic/stochastic sampling for a broader class of diffusion models is a promising direction. We propose two complementary frameworks for accelerating sample generation in pre-trained models: Conjugate Integrators and Splitting Integrators. Conjugate integrators generalize DDIM, mapping the reverse diffusion dynamics to a more amenable space for sampling. In contrast, splitting-based integrators, commonly used in molecular dynamics, reduce the numerical simulation error by cleverly alternating between numerical updates involving the data and auxiliary variables. After extensively studying these methods empirically and theoretically, we present a hybrid method that leads to the best-reported performance for diffusion models in augmented spaces. Applied to Phase Space Langevin Diffusion [Pandey & Mandt, 2023] on CIFAR-10, our deterministic and stochastic samplers achieve FID scores of 2.11 and 2.36 in only 100 network function evaluations (NFE) as compared to 2.57 and 2.63 for the best-performing baselines, respectively. Our code and model checkpoints will be made publicly available at \url{https://github.com/mandt-lab/PSLD}.