AI Paper Atlas

Research map / Video / Video generation

AdaCluster: Adaptive Query-Key Clustering for Sparse Attention in Video Generation

Haoyue Tan, Shengnan Wang, Yulin Qiao, Juncheng Zhang, Youhui Bai, Ping Gong, Zewen Jin, Cheng Li

CVPR 2026 · Video › Video generation · approach: Efficient video diffusion models · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add per-layer adaptive cluster counts instead of equal counts (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720PSNR: 29.007 -> 30.580Improved
add per-layer adaptive cluster counts instead of equal counts (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720SSIM: 0.724 -> 0.835Improved
add per-layer adaptive cluster counts instead of equal counts (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720LPIPS: 0.378 -> 0.203Improved
add TensorQuest cluster selection instead of mean-based selection (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720PSNR: 28.941 -> 30.580Improved
add TensorQuest cluster selection instead of mean-based selection (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720SSIM: 0.687 -> 0.835Improved
add TensorQuest cluster selection instead of mean-based selection (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720LPIPS: 0.410 -> 0.203Improved
add query normalization (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720PSNR: 29.56 -> 30.58Improved
add query normalization (ablation)AdaCluster on HunyuanVideo · HunyuanVideo 1280x720LPIPS: 0.317 -> 0.203Improved
raise sparsity and initial TopK (83.2%, TopK 32 setting)AdaCluster on HunyuanVideo · HunyuanVideo hyperparameter gridPSNR / speedup: 30.58 -> 28.82 PSNR, 1.68x -> 1.90xMixed
lower KV threshold to 3.5 (77.5% sparsity)AdaCluster on HunyuanVideo · HunyuanVideo hyperparameter gridPSNR / speedup: 30.58 -> 31.88 PSNR, 1.68x -> 1.69xMixed
use AdaCluster instead of SVG2AdaCluster · Wan2.1-14B on H100speedup: 1.61x -> 1.81xImproved
use AdaCluster instead of SpargeAttn and SVG2 at 28.3K tokensAdaCluster · HunyuanVideo efficiencyspeedup: 1.28x, 1.46x -> 1.53xImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy