AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

RMFlow: Refined Mean Flow by a Noise-Injection Step for Multimodal Generation

Yuhao Huang, Shih-Hsin Wang, Andrea L. Bertozzi, Bao Wang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Rectified flow straightening · 2 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 noise-injection refinement and likelihood loss to 1-NFE MeanFlowRMFlow · COCO FID-30KFID: 27.31 -> 18.91Improved
add noise-injection refinement and likelihood loss to 1-NFE MeanFlowRMFlow · COCO2017 CLIP scoreCLIP score: 0.273 -> 0.291Improved
add noise-injection refinement and likelihood lossRMFlow w/ contexts · QM9Molecule stability (%): 84.3 -> 93.2Improved
add context encoding (ablation)RMFlow · QM9Molecule stability (%): 90.1 -> 93.2Improved
add policy-gradient physical feedback (RLPF)RMFlow w/ contexts · QM9Molecule stability (%): 93.2 -> 93.5No clear effect
add noise-injection refinement to 1-NFE MeanFlowRMFlow · 1D Gaussian mixtureKL divergence: 0.8074 -> 0.2332Improved
1-NFE RMFlow vs 32-NFE MeanFlowRMFlow · 1D Gaussian mixtureKL divergence: 0.1017 vs 0.2332Hurt
add RMFlow refinement on dynamical systems (1-NFE)RMFlow · Lorenz w/o eventTV distance: 0.0469 -> 0.0332Improved
add RMFlow refinement (KL, Lorenz w/o event)RMFlow · Lorenz w/o eventKL divergence: 0.0109 -> 0.0059Improved
add likelihood loss weight lambda_1=1e-1 (vs 0) (ablation)RMFlow · 2D checkerboardKL divergence: 0.311 -> 0.163Improved
raise lambda_1 to 1e2 (ablation)RMFlow · 2D checkerboardKL divergence: 0.163 -> 0.425Hurt
raise lambda_1 to 1e1 (ablation)RMFlow · QM9Molecule stability (%): 93.2 -> 83.9Hurt

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