AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Mean Flow Distillation: Robust and Stable Distillation for Flow Matching Models

An Zhao, Shengyuan Zhang, Zhongjian Sun, Yixiang Zhou, Zejian Li, Ling Yang, Tianrun Chen, Lingyun Sun

ICML 2026 · Image generation › Fast diffusion sampling · approach: Flow matching distillation

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
match time-integrated mean flow instead of instantaneous score (VSD / Diff-Instruct)single-step student of OccFM · nuScenes 4D occupancy forecastingmIoU (%): 26.42 -> 27.25Improved
distill OccFM into one step with MFD versus teacher at 1 NFEOccFM · nuScenes 4D occupancy forecastingmIoU (%): 8.15 -> 27.25Improved
use MFD instead of AYF consistency distillationsingle-step student of OccFM · nuScenes 4D occupancy forecastingmIoU (%): 21.69 -> 27.25Improved
use MFD instead of DMD2/SenseFlow (adversarial) distillation4-step SANA 1.6B · text-to-image, FID (DINOv2)FID: 36.13 -> 33.28 (vs DMD2)Improved
use MFD instead of AYF4-step SANA 1.6B · text-to-imageFID (DINOv2): 43.68 -> 33.28Improved
reduce ODE solver step size to 0.001 (ablation)MFD · nuScenes 4D occupancy forecastingmIoU (%): 27.25 -> 25.91Hurt
increase ODE solver step size to 0.1 or 0.5 (ablation)MFD · nuScenes 4D occupancy forecastingmIoU (%): 27.25 -> 26.47 (0.1), 26.54 (0.5)Hurt
use moderate ODE step size 0.005 to 0.02 (ablation)MFD · nuScenes 4D occupancy forecastingmIoU (%): 27.14 (0.005), 27.25 (0.02)Improved
temporal integration in MFD instead of VSD single-timestep gradientMFD vs Diff-Instruct · text-to-image, SANA 1.6Bloss varianceImproved
reduce integration to single step (step size equals t-s) (ablation)MFD · variance/stability analysistraining stabilityHurt
MFD training cost versus DMD2SANA 1.6B distillation · training costpeak memory (GB): ~50 -> ~20Improved
MFD without adversarial or perceptual lossesMFD · text-to-image, SANA 1.6BPickScore / HPSv2: 0.2002 vs 0.2004 (DMD2); 0.2442 vs 0.2445 (SenseFlow)No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 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