Hacker News: thefox96https://news.ycombinator.com/user?id=thefox96Hacker News RSShttps://hnrss.org/hnrss v2.1.1Fri, 05 Jun 2026 21:07:11 +0000<![CDATA[New comment by thefox96 in "KVarN: Native vLLM backend for KV-cache quantization by Huawei"]]>it should be easy to do btw

]]>Thu, 04 Jun 2026 17:28:33 +0000https://news.ycombinator.com/item?id=48401814thefox96https://news.ycombinator.com/item?id=48401814https://news.ycombinator.com/item?id=48401814<![CDATA[New comment by thefox96 in "KVarN: Native vLLM backend for KV-cache quantization by Huawei"]]>Faster than Fp16, not better quality i guess

]]>Thu, 04 Jun 2026 17:02:26 +0000https://news.ycombinator.com/item?id=48401457thefox96https://news.ycombinator.com/item?id=48401457https://news.ycombinator.com/item?id=48401457