Dynamic Infilling Anchors for Format-Constrained Generation in Diffusion Large Language Models 事件

Dynamic Infilling Anchors for Format-Constrained Generation in Diffusion Large Language Models arXiv:2606.04535v1 Announce Type: new Abstract: Diffusion large language models (dLLMs) offer bidirectional attention and parallel generation, enabling them to exploit global context and naturally support format-constrained tasks like parseable JSON or reasoning templates. While straightforward fixed anchors can enforce such constraints, they often impose rigid spans, leading to truncated reasoning or