jdeschena/sdtt

Features

• Inference Acceleration - Accelerates language models via self-distillation through time.
• Language Diffusion Models - Fast LLMs via self-distillation through time.