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

(github.com)

66 points | by theanonymousone 3 hours ago ago

7 comments