Adaptive Spectral Feature Forecasting for Diffusion Sampling Acceleration

Excerpt

arXiv:2603.01623v2 Announce Type: replace-cross Abstract: Diffusion models have become the dominant tool for high-fidelity image and video generation, yet are critically bottlenecked by their inference speed due to the numerous iterative passes of Diffusion Transformers. To reduce the exhaustive compute, recent works resort to the feature caching and reusing scheme that skips network evaluations at selected diffusion steps by using cached features in previous steps. However, their preliminary de