<?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>Ppo on HJJJ的个人小站</title><link>https://hjjj.top/tags/ppo/</link><description>Recent content in Ppo on HJJJ的个人小站</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 19 Sep 2026 23:28:58 +0800</lastBuildDate><atom:link href="https://hjjj.top/tags/ppo/index.xml" rel="self" type="application/rss+xml"/><item><title>PPO 与 GRPO：从策略裁剪到组相对优势</title><link>https://hjjj.top/posts/computer-science/deep-learning/ppo-and-grpo/</link><pubDate>Fri, 11 Sep 2026 00:00:00 +0800</pubDate><guid>https://hjjj.top/posts/computer-science/deep-learning/ppo-and-grpo/</guid><description>从 PPO 的优势函数与概率比率出发，严谨理解 GRPO 如何用组内相对奖励训练语言模型</description></item></channel></rss>