verifiedhigh · 0.88updated 2026-10-03

ZipCache: Accurate and Efficient KV Cache Quantization with Salient Token Identification (Zhang et al., ICT CAS)

paper

Fields

sources/qlab-zipcache-paper

URL

https://arxiv.org/abs/2405.14256

Publisher

arXiv

Kind

paper

Published

2024-05-23

Accessed

2026-10-03

notes

Channel-separable tokenwise quantization + normalized attention score for saliency. Distinct from the unrelated DRAM/SSD 'ZipCache' system paper (2411.03174).

Notes

Channel-separable tokenwise quantization + normalized attention score for saliency. Distinct from the unrelated DRAM/SSD 'ZipCache' system paper (2411.03174).

Cross-references

This record references no other records.