Search papers, labs, and topics across Lattice.
The paper introduces SeaCache, a training-free caching schedule for accelerating diffusion model inference by reusing intermediate outputs. SeaCache addresses the limitations of existing caching strategies that rely on raw feature differences by incorporating spectral evolution awareness, which separates content and noise. By using a Spectral-Evolution-Aware (SEA) filter to estimate redundancy, SeaCache dynamically adapts to content while respecting spectral priors, achieving state-of-the-art latency-quality trade-offs across various visual generative models.
Ditch noisy feature distances: SeaCache uses a spectral filter to cache and reuse intermediate diffusion model outputs, slashing latency while maintaining image quality.
Diffusion models are a strong backbone for visual generation, but their inherently sequential denoising process leads to slow inference. Previous methods accelerate sampling by caching and reusing intermediate outputs based on feature distances between adjacent timesteps. However, existing caching strategies typically rely on raw feature differences that entangle content and noise. This design overlooks spectral evolution, where low-frequency structure appears early and high-frequency detail is refined later. We introduce Spectral-Evolution-Aware Cache (SeaCache), a training-free cache schedule that bases reuse decisions on a spectrally aligned representation. Through theoretical and empirical analysis, we derive a Spectral-Evolution-Aware (SEA) filter that preserves content-relevant components while suppressing noise. Employing SEA-filtered input features to estimate redundancy leads to dynamic schedules that adapt to content while respecting the spectral priors underlying the diffusion model. Extensive experiments on diverse visual generative models and the baselines show that SeaCache achieves state-of-the-art latency-quality trade-offs.