While state-of-the-art language models achieve impressive results through next-token prediction, they have inherent limitations such as the inability to revise already generated tokens. This has prompted exploration of alternative approaches such as discrete diffusion. However, masked diffusion, which has emerged as a popular choice due to its simplicity and effectiveness, reintroduces this inability to revise words. To overcome this, we generalize masked diffusion and derive the theoretical backbone of a family of general interpolating discrete diffusion (GIDD) processes offering greater flexibility in the design of the noising processes. Leveraging a novel diffusion ELBO, we achieve compute-matched state-of-the-art performance in diffusion language modeling. Exploiting GIDD's flexibility, we explore a hybrid approach combining masking and uniform noise, leading to improved sample quality and unlocking the ability for the model to correct its own mistakes, an area where autoregressive models notoriously have struggled. Our code and models are open-source:this https URL
View on arXiv@article{rütte2025_2503.04482,
title={ Generalized Interpolating Discrete Diffusion },
author={ Dimitri von Rütte and Janis Fluri and Yuhui Ding and Antonio Orvieto and Bernhard Schölkopf and Thomas Hofmann },
journal={arXiv preprint arXiv:2503.04482},
year={ 2025 }
}