<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>工程实践 on zzszmyf</title><link>https://zzszmyf.github.io/categories/%E5%B7%A5%E7%A8%8B%E5%AE%9E%E8%B7%B5/</link><description>Recent content in 工程实践 on zzszmyf</description><generator>Hugo</generator><language>zh-CN</language><copyright>© 2026 zzszmyf</copyright><lastBuildDate>Fri, 08 May 2026 15:00:00 +0800</lastBuildDate><atom:link href="https://zzszmyf.github.io/categories/%E5%B7%A5%E7%A8%8B%E5%AE%9E%E8%B7%B5/index.xml" rel="self" type="application/rss+xml"/><item><title>vLLM GPU 利用率持续 100% 排查记：一个 max_tokens 参数引发的性能陷阱</title><link>https://zzszmyf.github.io/notes/vllm-gpu-100-percent-max-tokens-trap/</link><pubDate>Fri, 08 May 2026 15:00:00 +0800</pubDate><guid>https://zzszmyf.github.io/notes/vllm-gpu-100-percent-max-tokens-trap/</guid><description/></item><item><title>用 kimi-cli 实现自动 Issue → Worktree → TDD → PR 的无人值守流水线</title><link>https://zzszmyf.github.io/notes/kimi-cli-auto-issue-pipeline/</link><pubDate>Tue, 05 May 2026 14:00:00 +0800</pubDate><guid>https://zzszmyf.github.io/notes/kimi-cli-auto-issue-pipeline/</guid><description/></item></channel></rss>