On the Error-Correcting Effects of Stochasticity in Discrete Diffusion

ArXi:2605.26582v1 Announce Type: cross Discrete diffusion models achieve strong performance in text and image generation, but their inference remains slow and must inherently balance sampling efficiency and sample quality. In this work, we present a systematic study of how the \emph{degree of stochasticity} in Marko transitions governs the sampling tradeoff. We show that highly deterministic transitions converge rapidly but suffer from error accumulation, while stochastic transitions converge slowly yet can achieve higher final sample quality.