TY  - RPRT
TI  - Revise, Don't Freeze: Sampler-Matched Training for Self-Correcting Masked Diffusion Language Models
AU  - Longxuan Yu
AU  - Shaorong Zhang
AU  - Yu Fu
AU  - Hui Liu
AU  - Yue Dong
AU  - Greg Ver Steeg
PY  - 2026
UR  - https://arxiv.org/abs/2606.01026
ID  - 2606.01026
ER  -