NeurIPS / Working Memory / Latent

Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time

作者：作者信息待提取 · 单位：Department of Computer Science, Rice University · 来源：NeurIPS · 发布日期：2024/03

工作记忆潜变量记忆
新标签打开 PDF
Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time figure
在线论文 PDF（可滚动查看）