GRKV: Global Regression for Training-Free KV Cache Compression in Long-Context LLMs 文章

ArXiv CS.CL2026-06-01NEWSen作者: Junjie Peng, You Wu, Haoyi Wu, Jialong Han, Xiaohua Xie, Kewei Tu, Jianhuang Lai

查看原文 →

GRKV: Global Regression for Training-Free KV Cache Compression in Long-Context LLMs · 相关技术

相关技术

post-eviction merging GRKV KV Cache span-based retention large language models key-value (KV) cache