"KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache."

Zirui Liu et al. (2024)

Details and statistics

DOI: 10.48550/ARXIV.2402.02750

access: open

type: Informal or Other Publication

metadata version: 2024-04-30

a service of  Schloss Dagstuhl - Leibniz Center for Informatics