04:00
2026-05-27
arxiv.org
artificial-intelligence
Targeted Remasking: Replacing Token Editing with Token-to-Mask Refinement in Discrete Diffusion Language Models
Researchers have developed Token-to-Mask (T2M) remasking, a training-free replacement for Token-to-Token (T2T) editing in discrete masked diffusion language models like LLaDA. The method resets suspec…