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

by theanonymousone | View on Hacker News