AI Paper Atlas

Research map / Video / Video generation

Attention Surgery: An Efficient Recipe to Linearize Your Video Diffusion Transformer

Mohsen Ghafoorian, Denis Korzhenkov, Amirhossein Habibian

CVPR 2026 · Video › Video generation · approach: Efficient video diffusion models · 10 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 attention distillation to 15 linear-attention blocks (ablation)Attention Surgery · VBench (ablation)Total: 59.7 -> 78.9Improved
add attention distillation to 20xR8 hybrid (ablation)Attention Surgery · VBench (ablation)Total: 77.3 -> 80.0Improved
use heterogeneous block-rate selection instead of homogeneous (20xR8) (ablation)Attention Surgery · VBench (ablation)Total: 80.0 -> 80.9Improved
use heterogeneous block-rate selection instead of homogeneous (15xR4) (ablation)Attention Surgery · VBench (ablation)Total: 81.0 -> 81.9Improved
use heterogeneous block-rate selection instead of homogeneous (20xR4) (ablation)Attention Surgery · VBench (ablation)Total: 80.9 -> 81.1Improved
use heterogeneous block-rate selection instead of homogeneous (25xR4) (ablation)Attention Surgery · VBench (ablation)Total: 79.9 -> 80.2Improved
value distillation instead of attention distillation (20xR8) (ablation)Attention Surgery · VBench (ablation)Total: 79.5 -> 80.0Improved
value distillation instead of attention distillation (15xR8) (ablation)Attention Surgery · VBench (ablation)Total: 81.2 -> 80.1Hurt
increase phi complexity from poly degree 2 / 2 layers to degree 6 / 3 layers (ablation)Attention Surgery · VBench (ablation)Total (10xR8): 82.1 -> 82.1No clear effect
replace original Wan2.1 attention with 15xR2 hybrid attention surgeryWan2.1 1.3B · VBenchTotal: 83.10 -> 83.21No clear effect
replace original Wan2.1 with 15xR2 hybridWan2.1 1.3B · VBench-2.0Total: 56.0 -> 55.1Hurt
replace original Wan2.1 with 15xR2 hybridWan2.1 1.3B · Human preference studypreference: 31.0 vs 29.3 preferred; 39.7 no preferenceNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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