RECAP: Reciprocal Sparse Attention and Hierarchical Caching for Efficient Video Diffusion
Diffusion transformers generate high-quality videos but remain computationally expensive.
ProofPaper ↗
Key points
- Sparse attention and temporal caching reduce computation within and across denoising steps, respectively, offering a natural composition.
- These observations motivate RECAP, a training-free method that couples sparse attention with hierarchical caching through reciprocal correction, adaptively coordinated by online error feedback.
- Cached dense refinement corrects omitted interactions in current sparse estimates, while fresh sparse statistics track input changes and regulate the aging correction.
- We evaluate RECAP on multiple video diffusion models against sparse-attention methods, caching methods, and their direct compositions.
Sources (1)
- [1]RECAP: Reciprocal Sparse Attention and Hierarchical Caching for Efficient Video DiffusionarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 12:03 PM
Diffusion transformers generate high-quality videos but remain computationally expensive.
Sparse attention and temporal caching reduce computation within and across denoising steps, respectively, offering a natural composition.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 19, 2026TamedFlow: Trajectory-Aware Distillation\\for Teacher-Assisted Diffusion Sampling
- Oct 19, 2026Any 3D Scene is Worth 1K Tokens: 3D-Grounded Representation for Scene Generation at Scale
- Oct 19, 2026SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation
- Oct 19, 2026SpecKCast: Spectral Kalman Forecasting for Diffusion Acceleration
- Oct 19, 2026RoAR-VAE: Improving Latent Diffusability through Robust Prediction
- Oct 18, 2026AffordanceWAM: Affordance-Aware Joint World-Action Modeling for Robot Manipulation