Dystruct: Dynamically Structured Diffusion Language Model Decoding via Bayesian Inference

ArXi:2605.09820v1 Announce Type: new Diffusion language models (DLMs) have recently emerged as a promising alternative to autoregressive models, primarily due to their ability to enable parallel decoding. Despite this advantage, most existing DLMs rely on a fixed generation length specified prior to decoding, which restricts their flexibility in real-world applications. While a few recent works attempt to flexible-length generation, they typically suffer from notable limitations: some require costly re.