Unlocking Lossless Speedups in LLMs via Discrete Diffusion

This project unlocks lossless speedups for large language models using discrete diffusion. To adopt it, first manually verify compatibility with your stack, the