Search for a command to run...
Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing