1 paper
Yangjia Hu, Haodong Wang, Zicong Hong +8
4-bit quantization significantly reduces the memory footprint and accelerates the inference of large language models (LLMs). However, its limited bit-width representation struggles…