KVarN: new KV-cache quant from Huawei. 3–5× KV cache compression with actual speed-up instead of slow-down, and unlike TurboQuant it holds up on reasoning (Apache 2.0, vLLM single flag) | AI.News
KVarN: new KV-cache quant from Huawei. 3–5× KV cache compression with actual speed-up instead of slow-down, and unlike TurboQuant it holds up on reasoning (Apache 2.0, vLLM single flag)