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

(github.com)

51 points | by theanonymousone 2 hours ago ago

7 comments