Rethinking Token-wise Feature Caching: Accelerating Diffusion Transformers with Dual Feature Caching
This work addresses computational bottlenecks in diffusion-based image/video generation models, offering a more efficient acceleration method, though it appears incremental as it builds on existing caching approaches.
The paper challenges the effectiveness of token-wise feature caching for accelerating Diffusion Transformers (DiT), showing that computing 'important' tokens consistently is unnecessary and selection methods can be inferior to random selection, and proposes DuCa, a dual caching strategy with random token selection that achieves significant improvements over previous methods across DiT, PixArt, FLUX, and OpenSora.
Diffusion Transformers (DiT) have become the dominant methods in image and video generation yet still suffer substantial computational costs. As an effective approach for DiT acceleration, feature caching methods are designed to cache the features of DiT in previous timesteps and reuse them in the next timesteps, allowing us to skip the computation in the next timesteps. Among them, token-wise feature caching has been introduced to perform different caching ratios for different tokens in DiTs, aiming to skip the computation for unimportant tokens while still computing the important ones. In this paper, we propose to carefully check the effectiveness in token-wise feature caching with the following two questions: (1) Is it really necessary to compute the so-called "important" tokens in each step? (2) Are so-called important tokens really important? Surprisingly, this paper gives some counter-intuition answers, demonstrating that consistently computing the selected ``important tokens'' in all steps is not necessary. The selection of the so-called ``important tokens'' is often ineffective, and even sometimes shows inferior performance than random selection. Based on these observations, this paper introduces dual feature caching referred to as DuCa, which performs aggressive caching strategy and conservative caching strategy iteratively and selects the tokens for computing randomly. Extensive experimental results demonstrate the effectiveness of our method in DiT, PixArt, FLUX, and OpenSora, demonstrating significant improvements than the previous token-wise feature caching.