CriticalKV: Optimizing KV Cache Eviction from an Output Perturbation Perspective 文章

ArXiv CS.CL2026-05-29NEWSen作者: Yuan Feng, Junlin Lv, Haoyu Guo, Yukun Cao, S Kevin Zhou, Xike Xie

查看原文 →

CriticalKV: Optimizing KV Cache Eviction from an Output Perturbation Perspective · 相关技术

相关技术

perturbation-constrained selection algorithm attention output perturbation KV Cache self-attention Transformer architecture