Bell-shaped time sampling — drawing the corruption level near t=0.5 — accelerates masked diffusion language model training by up to ~4× without changing the final loss.
Demystifying MaskGIT sampler and beyond: Adaptive order selection in masked diffusion
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Understanding and Accelerating the Training of Masked Diffusion Language Models
Bell-shaped time sampling — drawing the corruption level near t=0.5 — accelerates masked diffusion language model training by up to ~4× without changing the final loss.