AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

QVGen: Pushing the Limit of Quantized Video Generative Models

Yushi Huang, Ruihao Gong, Jing Liu, Yifu Ding, Chengtao Lv, Haotong Qin, Jun Zhang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 16 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 auxiliary module Phi to naive KD-based QAT (ablation)Naive QAT, W4A4 Wan 1.3B · VBenchImaging Quality: 60.40 -> 63.41Improved
add rank-decay to remove Phi (ablation)Naive + Phi, W4A4 Wan 1.3B · VBenchImaging Quality: 63.41 -> 63.08No clear effect
add rank-decay to remove Phi (ablation)Naive + Phi, W4A4 Wan 1.3B · VBenchOverall Consistency: 22.98 -> 23.01Improved
use shrinking ratio 1/2 instead of 1 (decaying all Phi at once) (ablation)QVGen W4A4 Wan 1.3B · VBenchAesthetic Quality: 52.48 -> 54.67Improved
use initial rank r=32 instead of 8 (ablation)QVGen W4A4 Wan 1.3B · VBenchDynamic Degree: 74.62 -> 77.78Improved
increase initial rank from 32 to 64 (ablation)QVGen W4A4 Wan 1.3B · VBenchDynamic Degree: 77.78 -> 76.74Hurt
use rank-decay instead of linear decay (ablation)Phi decay, W4A4 Wan 1.3B · VBenchImaging Quality: 60.82 -> 63.08Improved
use rank-decay instead of Sparse with MaskLLM (ablation)Phi decay, W4A4 Wan 1.3B · VBenchImaging Quality: 61.36 -> 63.08Improved
use QVGen instead of EfficientDM on 3-bit CogVideoX-2BW3A3 CogVideoX-2B · VBenchDynamic Degree: 28.61 -> 53.89Improved
use QVGen instead of Q-DM on 4-bit Wan 1.3BW4A4 Wan 1.3B · VBenchAesthetic Quality: 52.50 -> 54.67Improved
mix half high-motion and half low-motion training videos instead of high-motion only (ablation)W4A4 QVGen CogVideoX-2B · VBenchDynamic Degree: 68.21 -> 67.08Mixed
use QVGen on larger Wan 14B at 3 bits instead of 4 bitsQVGen Wan 14B · VBenchAesthetic Quality: 59.41 -> 29.73Hurt

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