AutoAWQ is an easy-to-use package for 4-bit quantized models that speeds up LLM inference by 3x and...