<?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>Distributed-Training on LLM Infra Tutorial</title><link>https://mzf666.github.io/llm-infra/en/tags/distributed-training/</link><description>Recent content in Distributed-Training on LLM Infra Tutorial</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Tue, 17 Mar 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://mzf666.github.io/llm-infra/en/tags/distributed-training/index.xml" rel="self" type="application/rss+xml"/><item><title>The Landscape of Distributed Parallelism Strategies</title><link>https://mzf666.github.io/llm-infra/en/posts/02-parallel-strategies/</link><pubDate>Mon, 16 Mar 2026 00:00:00 +0000</pubDate><guid>https://mzf666.github.io/llm-infra/en/posts/02-parallel-strategies/</guid><description>From DDP to hybrid parallelism — a systematic guide to every parallelism strategy in large model training.</description></item><item><title>Introduction to RLHF System Design</title><link>https://mzf666.github.io/llm-infra/en/posts/04-rlhf-system/</link><pubDate>Tue, 17 Mar 2026 00:00:00 +0000</pubDate><guid>https://mzf666.github.io/llm-infra/en/posts/04-rlhf-system/</guid><description>From the four-model RLHF architecture to verl&amp;rsquo;s system design — understanding why RLHF is fundamentally a systems problem.</description></item></channel></rss>