KVarN: Native vLLM backend for KV-cache quantization by Huawei

66 points | by theanonymousone 3 hours ago

7 comments