<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
    <channel>
        <title>Reading-Note/Distributed-Training - 标签 - vllbc02&#39;s blogs</title>
        <link>https://blog.vllbc.top/tags/reading-note/distributed-training/</link>
        <description>Reading-Note/Distributed-Training - 标签 - vllbc02&#39;s blogs</description>
        <generator>Hugo -- gohugo.io</generator><language>zh-CN</language><copyright>This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.</copyright><lastBuildDate>Wed, 26 Aug 2026 00:00:00 &#43;0000</lastBuildDate><atom:link href="https://blog.vllbc.top/tags/reading-note/distributed-training/" rel="self" type="application/rss+xml" /><item>
    <title>Ultra-Scale Playbook · High-Level Overview 速查</title>
    <link>https://blog.vllbc.top/ultrascale-playbook-high-level-overview/</link>
    <pubDate>Wed, 26 Aug 2026 00:00:00 &#43;0000</pubDate>
    <author>Nouamane Tazi, Ferdinand Mom, Haojun Zhao, Mohamed Mekkouri, Leandro von Werra, Thomas Wolf (Hugging Face nanotron)</author>
    <guid>https://blog.vllbc.top/ultrascale-playbook-high-level-overview/</guid>
    <description><![CDATA[Hugging Face nanotron 团队基于 4100+ 实验、512 GPU 集群的分布式训练方法论。
原笔记 [[UltraScale Playbook 笔记]] 是完整教材；本文聚焦 high-level overview，
提供 5D 并行速查表 + 选型决策树 + 集合通信原语速查 + 显存四件套对照，
目标是 30 秒抓住主线、5 分钟能查、能选。]]></description>
</item>
</channel>
</rss>
