KVarN: Native vLLM backend for KV-cache quantization by Huaweigithub.com
70 points by theanonymousone 3 hours ago | 7 comments