AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency

Kaiwen Zheng, Yuji Wang, Qianli Ma, Huayu Chen, Jintao Zhang, Yogesh Balaji, Jianfei Chen, Ming-Yu Liu, Jun Zhu, Qinsheng Zhang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 65 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 DMD score-distillation regularizer to sCM (rCM vs pure sCM) (ablation)sCM · T2I/T2V distillationvisual qualityImproved
rCM versus DMD2 on image qualityrCM (Cosmos-Predict2 2B, 4 steps) · GenEval (Cosmos-Predict2 T2I)GenEval overall: 0.80 -> 0.81No clear effect
rCM versus DMD2 on video qualityrCM (Wan2.1 1.3B, 4 steps) · VBench total score (Wan2.1 T2V, 832x480x81)VBench total score: 84.56 -> 84.43No clear effect
rCM versus DMD2 on diversityrCM (Wan2.1) · Video diversitysample diversityImproved
scale rCM student from 0.6B to 14BrCM · GenEval (Cosmos-Predict2 T2I)GenEval overall: 0.79 -> 0.83Improved
reduce sampling steps from 4 to 1rCM (Cosmos-Predict2 14B) · GenEval (Cosmos-Predict2 T2I)GenEval overall: 0.83 -> 0.82Hurt
reduce sampling steps from 4 to 1rCM (Wan2.1 14B) · VBench total score (Wan2.1 T2V, 832x480x81)VBench total score: 84.92 -> 83.02Hurt
reduce sampling steps from 4 to 2rCM (Wan2.1 14B) · VBench total score (Wan2.1 T2V, 832x480x81)VBench total score: 84.92 -> 85.05No clear effect
distill with rCM versus teacher on videorCM (Wan2.1 14B) · VBench total score (Wan2.1 T2V, 832x480x81)VBench total score: 83.58 -> 84.92Improved
increase DMD weight lambda (ablation)rCM (Wan2.1 1.3B) · Video quality vs diversityquality and diversity: larger lambda: better quality, less diversityMixed
use continuous-time consistency trajectory model (sCTM/MeanFlow) instead of sCM (ablation)sCM · T2I distillationquality and diversityHurt
use SiD instead of DMD as score regularizer (ablation)rCM · T2I/T2VqualityNo clear effect

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