Fetching the paper…

Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing · Around