LLM KV缓存管理：安全威胁与效率创新

报道来源 [4]

1. arXiv cs.AI TIER_1 English(EN) · Peng Xie · 2026-07-24 04:00

   KV-Cache 驱逐的随机设计错误证书

   arXiv:2607.21475v1 Announce Type: cross Abstract: Deterministic KV-cache eviction keeps the top-$k$ tokens under an importance score and deletes the rest. We prove that this design cannot know what it destroyed: evicted values can be altered so that everything the serving system …

2. Hugging Face Daily Papers TIER_1 English(EN) · 2026-07-23 16:16

   Error Certificates for KV-Cache Eviction via Randomized Design

   Deterministic KV-cache eviction keeps the top-$k$ tokens under an importance score and deletes the rest. We prove that this design cannot know what it destroyed: evicted values can be altered so that everything the serving system retains is unchanged while the true attention-outp…

3. arXiv cs.AI TIER_1 English(EN) · Yichi Zhang, Zhiqi Wang, Huan Zhang, Yuchen Yang · 2026-07-23 04:00

   HijackKV：位置无关 KV 缓存重用中的新威胁

   arXiv:2607.19957v1 Announce Type: cross Abstract: Key-Value (KV) cache reduces inference latency in large language models (LLMs). Traditional prefix-based reuse has low cache hit rates across inference requests because it requires exact token and position matches. To improve effi…

4. arXiv cs.AI TIER_1 English(EN) · Soumia Bouyahiaoui, Manel Kara laouar, Aicha Boutorh, Mohamed Hadj Ameur · 2026-07-21 04:00

   SelKV：具有逐令牌合并或丢弃和注意力补偿的选择性KV缓存合并

   arXiv:2607.16213v1 Announce Type: new Abstract: Large Language Models (LLMs) generate text autoregressively, relying on a key-value (KV) cache whose memory footprint grows linearly with context length, creating a major bottleneck. Recent compression methods mitigate this cost via…