arXiv:2604.08557v2 Announce Type: replace
Abstract: Safety alignment in diffusion language models (dLLMs) relies on a single load-bearing assumption: that committed tokens are permanent. We show that violating this assumption, by re-masking committed …
arXiv:2604.08557v1 Announce Type: new
Abstract: Diffusion-based language models (dLLMs) generate text by iteratively denoising masked token sequences. We show that their safety alignment rests on a single fragile assumption: that the denoising schedul…
arXiv:2604.08556v1 Announce Type: new
Abstract: What exactly do efficient sequence models gain over simple temporal averaging? We use exponential moving average (EMA) traces, the simplest recurrent context (no gating, no content-based retrieval), as a…