<?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>RRSI on Simon Sun</title><link>https://fatflowers.github.io/zh/tags/rrsi/</link><description>Recent content in RRSI on Simon Sun</description><image><title>Simon Sun</title><url>https://fatflowers.github.io/images/social-card-v2.jpg</url><link>https://fatflowers.github.io/images/social-card-v2.jpg</link></image><generator>Hugo -- 0.155.3</generator><language>zh</language><lastBuildDate>Wed, 23 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://fatflowers.github.io/zh/tags/rrsi/index.xml" rel="self" type="application/rss+xml"/><item><title>RRSI 论文解读：Agent 会自己改 Harness，也会学会刷题</title><link>https://fatflowers.github.io/zh/posts/original-tech/rrsi-agent-harness/</link><pubDate>Wed, 23 Sep 2026 00:00:00 +0000</pubDate><guid>https://fatflowers.github.io/zh/posts/original-tech/rrsi-agent-harness/</guid><description>解读 RRSI 如何约束 Agent Harness 的自动演化，让改进更可能迁移到没见过的任务。</description></item></channel></rss>