Distilling ODE Solvers of Diffusion Models into Smaller Steps
Sanghwan Kim,Hao Tang,Fisher Yu
DOI: https://doi.org/10.48550/arXiv.2309.16421
2024-03-27
Abstract:Abstract Diffusion models have recently gained prominence as a novel category of generative models. Despite their success, these models face a notable drawback in terms of slow sampling speeds, requiring a high number of function evaluations (NFE) in the order of hundreds or thousands. In response, both learning-free and learning-based sampling strategies have been explored to expedite the sampling process. Learning-free sampling employs various ordinary differential equation (ODE) solvers based on the formulation of diffusion ODEs. However, it encounters challenges in faithfully tracking the true sampling trajectory, particularly for small NFE. Conversely, learning-based sampling methods, such as knowledge distillation, demand extensive additional training, limiting their practical applicability. To overcome these limitations, we introduce Distilled-ODE solvers (D-ODE solvers), a straightforward distillation approach grounded in ODE solver formulations. Our method seamlessly integrates the strengths of both learning-free and learning-based sampling. D-ODE solvers are constructed by introducing a single parameter adjustment to existing ODE solvers. Furthermore, we optimize D-ODE solvers with smaller steps using knowledge distillation from ODE solvers with larger steps across a batch of samples. Comprehensive experiments demonstrate the superior performance of D-ODE solvers compared to existing ODE solvers, including DDIM, PNDM, DPM-Solver, DEIS, and EDM, particularly in scenarios with fewer NFE. Notably, our method incurs negligible computational overhead compared to previous distillation techniques, facilitating straightforward and rapid integration with existing samplers. Qualitative analysis reveals that D-ODE solvers not only enhance image quality but also faithfully follow the target ODE trajectory.
Computer Vision and Pattern Recognition