iCATS: Fast Video Generation via Interaction-Aware Sparse Attention and Timestep-Adaptive Sparsity
cs.CV
Submitted: 2026-10-08
Updated: 2026-10-08
License: http://creativecommons.org/licenses/by/4.0/
The gist: Training-free sparse attention offers a practical acceleration solution to Diffusion Transformers (DiTs) via reducing computations without fine-tuning.
Terminology
Abstract
Training-free sparse attention offers a practical acceleration solution to Diffusion Transformers (DiTs) via reducing computations without fine-tuning. It typically involves estimating the importance of query-key regions and deriving sparse masks to compute only the important candidates, which inevitably introduces approximation errors that may degrade generation quality. To better balance the efficiency-quality trade-off, we propose iCATS, integrating improved importance estimation and sparse mask construction with an efficient hardware execution strategy. Specifically, for importance estimation, unlike previous works that perform independent clustering over query and key tokens based on feature similarity to estimate attention scores, iCATS demonstrates that clustering based on query-key dot-product interactions is more accurate and further reformulates this objective as a simple quadratic form for low-cost computation. For sparse mask construction, instead of using a fixed top-p rule, we observe that tolerance to sparse approximation errors varies across denoising timesteps and therefore introduce an SNR-guided sparsity schedule to adjust sparsity dynamically, leading to higher accuracy. Finally, for hardware execution, we devise a tail-merging strategy to reduce padding overhead caused by irregular cluster sizes, improving GPU kernel utilization. Extensive experiments show that iCATS achieves 2.03 times acceleration with 31.017 dB PSNR on HunyuanVideo-T2V-13B and 1.55 times acceleration with 29.301 dB PSNR on Wan2.1-T2V-14B, delivering a state-of-the-art efficiency-quality trade-off.
Sources
- Scalable Diffusion Models with Transformers
- Wan: Open and Advanced Large-Scale Video Generative Models
- HunyuanVideo: A Systematic Framework For Large Video Generative Models
- CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer
- Seedance 1.0: Exploring the Boundaries of Video Generation Models
- RainFusion: Adaptive Video Generation Acceleration via Multi-Dimensional Visual Redundancy
- RainFusion2.0: Temporal-Spatial Awareness and Hardware-Efficient Block-wise Sparse Attention
- Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
- Radial Attention: $O(n\log n)$ Sparse Attention with Energy Decay for Long Video Generation
- Rectified SpaAttn: Revisiting Attention Sparsity for Efficient Video Generation
- SLA2: Sparse-Linear Attention with Learnable Routing and Quantization-aware Training
- Flow Matching for Generative Modeling
- FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning
- DSV: Exploiting Dynamic Sparsity to Accelerate Large-Scale Video DiT Training
- VBench-2.0: Advancing Video Generation Benchmark Suite for Intrinsic Faithfulness
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models