Apple Machine Learning Research

Accelerating LLM Inference on NVIDIA GPUs with ReDrafter - Apple Machine Learning Research

Accelerating LLM inference is an important ML research problem, as auto-regressive token generation is computationally expensive and…