Tag
1 open-source project filed under this tag.
LMCache is a KV cache management layer that slashes LLM inference latency by reusing computed attention states across requests, sessions, and GPU clusters.