<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>VLLM on Adeline</title><link>https://adeline-ly-blog.pages.dev/tags/vllm/</link><description>Recent content in VLLM on Adeline</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Fri, 31 Jul 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://adeline-ly-blog.pages.dev/tags/vllm/index.xml" rel="self" type="application/rss+xml"/><item><title>vLLM 如何管理混合架构模型的 KV Cache：从 Kimi Linear / Kimi K3 说起</title><link>https://adeline-ly-blog.pages.dev/posts/vllm-hybrid-kvcache/</link><pubDate>Fri, 31 Jul 2026 00:00:00 +0000</pubDate><guid>https://adeline-ly-blog.pages.dev/posts/vllm-hybrid-kvcache/</guid><description>以 Kimi Linear 与 Kimi K3 为例，梳理 vLLM 面对线性注意力 + 全注意力混合架构时，KV cache 如何组织、分配与复用。</description></item></channel></rss>