<?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>LLaMA on GopherDing's Blog</title><link>/tags/llama/</link><description>Recent content in LLaMA on GopherDing's Blog</description><generator>Hugo -- gohugo.io</generator><language>zh</language><lastBuildDate>Fri, 10 Apr 2026 15:55:23 +0800</lastBuildDate><atom:link href="/tags/llama/index.xml" rel="self" type="application/rss+xml"/><item><title>速通 AI（五）：现代大模型架构——LLaMA 系列</title><link>/posts/tech/%E9%80%9F%E9%80%9Aai-05-llama%E6%9E%B6%E6%9E%84/</link><pubDate>Fri, 10 Apr 2026 15:55:23 +0800</pubDate><guid>/posts/tech/%E9%80%9F%E9%80%9Aai-05-llama%E6%9E%B6%E6%9E%84/</guid><description>「速通 AI」系列第五篇：深入理解 LLaMA 架构的 5 大关键改进——RMSNorm、RoPE、SwiGLU、KV Cache、GQA。</description></item></channel></rss>