AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Don't Settle Too Early: Self-Reflective Remasking for Diffusion Language Models

Zemin Huang, Yuhang Wang, Zhiyang Chen, Guo-Jun Qi

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding

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 Remask SFT (train on incorrect-token noise with remask labels) instead of vanilla SFT (ablation)RemeDi · MATH-500accuracy: 40.1 -> 42.7Improved
add Remask SFT instead of vanilla SFT (ablation)RemeDi · HumanEvalpass@1: 48.2 -> 50.0Improved
add Remask SFT instead of vanilla SFT (ablation)RemeDi · GSM8K and MBPPaccuracy: GSM8K 83.1 -> 83.6; MBPP 43.4 -> 44.0Improved
add Remask RL after Remask SFT (ablation)RemeDi · AlpacaEvalwin rate: 12.5 -> 24.8Improved
add Remask RL after Remask SFT (ablation)RemeDi · GSM8K, MATH, HumanEvalaccuracy: GSM8K 86.3 -> 89.1; MATH 51.4 -> 52.9; HumanEval 71.3 -> 73.2Improved
add Remask RL after Remask SFT (ablation)RemeDi · GPQAaccuracy: 32.6 -> 29.5Hurt
use Remask RL instead of LLaDOU RL on LLaDA (ablation)GSM8K RL training · GSM8K pass@1, 200 stepspass@1: 82.35% -> 83.33% at step 200; 77.58% -> 80.00% at step 50Improved
add remasking-enabled training on LLaDA (RemeDi vs LLaDA)LLaDA-8B-Instruct · HumanEvalpass@1: 45.7 -> 73.2Improved
add inference-only random remasking (ReMDM) to LLaDALLaDA · MATHaccuracy: 38.9 -> 38.5No clear effect
harder problems on MATH-500 (level 1 to level 5)RemeDi · MATH-500remasking frequency per block: 9.13 -> 13.95 tokens per blockImproved

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