Submitted by Ligong Han 4 S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation Red Hat AI 6 1