AI RESEARCH
Variational Autoencoding Discrete Diffusion with Enhanced Dimensional Correlations Modeling
arXiv CS.LG
•
ArXi:2505.17384v2 Announce Type: replace Discrete diffusion models have recently shown great promise for modeling complex discrete data, with masked diffusion models (MDMs) offering a compelling trade-off between quality and generation speed. MDMs denoise by progressively unmasking multiple dimensions from an all-masked input, but their performance can degrade when using few denoising steps due to limited modeling of inter-dimensional dependencies.