Skip to contentSkip to filters

We use cookies for analytics.

Optimal Inference Schedules for Masked Diffusion Models

Jerry Li

University of Washington

Recording

40 min

Abstract

Masked diffusion language models can reveal tokens out of order and in parallel, potentially reducing sequential inference costs. This talk asks how much parallelism is possible before distributional accuracy degrades. It gives an exact expression for expected divergence under arbitrary data distributions and unmasking schedules, exposing a connection with classical approximation of one-variable functions.

Topics