Which Heads Matter for Reasoning? RL-Guided KV Cache Compression 文章

ArXiv CS.CL2026-05-28NEWSen作者: Wenjie Du, Li Jiang, Keda Tao, Xue Liu, Huan Wang

详细信息

来源站点
ArXiv CS.CL
作者
Wenjie Du, Li Jiang, Keda Tao, Xue Liu, Huan Wang
文章类型
NEWS
语言
en
发布日期
2026-05-28

摘要

arXiv:2510.08525v3 Announce Type: replace Abstract: Reasoning large language models exhibit complex reasoning behaviors via extended chain-of-thought generation that are highly fragile to information loss during decoding, creating critical challenges for KV cache compression. Existing token-dropping methods directly disrupt reasoning chains by removing intermediate steps, while head-reallocation methods, designed for retrieval tasks, fail to preserve the heads essential for generative reasoning. However, no existing method can identify which attention heads genuinely maintain reasoning consistency and control generation termination. To address this, we propose RLKV, which uses reinforcement learning as a probe to discover which heads contribute to reasoning quality by directly optimizing their cache usage against actual generation outcomes.

相关事件

暂无数据

相关公司

暂无数据

相关人物

暂无数据

相关产品

暂无数据