AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Bringing Stability to Diffusion: Decomposing and Reducing Variance of Training Masked Diffusion Models

Mengni Jia, Mengyu Zhou, Yihao Liu, xiaoxi jiang, guanjunjiang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Fast diffusion ODE solvers · 1 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 P-POTS+MIRROR to standard MDM trainingLLaDA-8B-Instruct · GSM8Kaccuracy (%): 52.51 -> 60.53Improved
add P-POTS+MIRROR to standard trainingLLaDA-8B-Instruct · HiTabaccuracy (%): 58.68 -> 67.10Improved
add P-POTS+MIRROR to standard trainingLLaDA-8B-Instruct · OpenScienceaccuracy (%): 45.52 -> 52.53Improved
add P-POTS+MIRROR to standard trainingMMaDA-8B-MixCoT · text-to-image-2MCLIP score: 31.36 -> 34.59Improved
add MIRROR on top of P-POTS (ablation)P-POTS · GSM8Kaccuracy (%): 58.58 -> 60.53Improved
add MIRROR on top of P-POTS (ablation)P-POTS · OpenScienceaccuracy (%): 46.80 -> 52.53Improved
add P-POTS on top of MIRROR (ablation)MIRROR · HiTabaccuracy (%): 64.48 -> 67.10Improved
add MIRROR alone to standard training (ablation)LLaDA-8B-Instruct · GSM8Kaccuracy (%): 52.51 -> 53.70Improved
add P-POTS alone to standard training (ablation)MMaDA-8B-MixCoT · text-to-image-2MCLIP score: 31.36 -> 31.65No clear effect
add MIRROR alone to standard training (ablation)MMaDA-8B-MixCoT · text-to-image-2MCLIP score: 31.36 -> 34.07Improved
use P-POTS+MIRROR instead of clipped noise scheduleLLaDA-8B-Instruct · GSM8Kaccuracy (%): 56.18 -> 60.53Improved
add stratified t-sampling (StraTS) to standard training (ablation)LLaDA-8B-Instruct · GSM8K (seed 42)accuracy (%): 50.64 -> 55.57Improved

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