1
Grounded Cache Routing for Retrieval-Augmented Generation: When Is It Safe to Reuse an Answer?
探讨RAG系统中缓存复用安全性的新方法,提出Grounded Cache Routing机制,优化成本与延迟。
arXiv:2605.27494v1 Announce Type: cross Abstract: Modern retrieval-augmented generation(RAG) deployments increasingly rely on caching to reduce token …