Skip to content

M-ANT: Efficient Low-bit Group Quantization for LLMs via Mathematically Adaptive Numerical Type.

Weiming Hu, Haoyan Zhang, Cong Guo, Yu Feng, Renyang Guan, Zhendong Hua, Zihan Liu, Yue Guan, Minyi Guo, Jingwen Leng

VenueA*HPCA
Year2025
ProceedingsHPCA

Browse the full HPCA paper archive.