<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
    <channel>
        <title>RL - 标签 - vllbc02&#39;s blogs</title>
        <link>https://blog.vllbc.top/tags/rl/</link>
        <description>RL - 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>Thu, 23 Jul 2026 00:00:00 &#43;0000</lastBuildDate><atom:link href="https://blog.vllbc.top/tags/rl/" rel="self" type="application/rss+xml" /><item>
    <title>Review：自进化Agent的RL Memory与Skill方法批判</title>
    <link>https://blog.vllbc.top/review%E8%87%AA%E8%BF%9B%E5%8C%96agent%E7%9A%84rl-memory%E4%B8%8Eskill%E6%96%B9%E6%B3%95%E6%89%B9%E5%88%A4/</link>
    <pubDate>Thu, 23 Jul 2026 00:00:00 &#43;0000</pubDate>
    <author>vllbc</author>
    <guid>https://blog.vllbc.top/review%E8%87%AA%E8%BF%9B%E5%8C%96agent%E7%9A%84rl-memory%E4%B8%8Eskill%E6%96%B9%E6%B3%95%E6%89%B9%E5%88%A4/</guid>
    <description><![CDATA[<blockquote>
<p>一篇对当前 “用 RL 提升 memory / skill refinement”
这条线的批判性综述。作者把现状分成 4 类：SkillRL、MemRL、纯 prompt
memory、search-augmented RL。本笔记整理作者的核心观点,并与 vault
内已有的相关论文笔记交叉链接。</p>]]></description>
</item>
<item>
    <title>RL 面试题 2026</title>
    <link>https://blog.vllbc.top/rl%E9%9D%A2%E8%AF%95%E9%A2%982026/</link>
    <pubDate>Tue, 30 Jun 2026 00:00:00 &#43;0000</pubDate>
    <author>vllbc</author>
    <guid>https://blog.vllbc.top/rl%E9%9D%A2%E8%AF%95%E9%A2%982026/</guid>
    <description><![CDATA[<blockquote>
<p>来源: 整理自 <a
href="https://x.com/sheriyuo/status/2063295181131247674"><span
class="citation" data-cites="sheriyuo">@sheriyuo</span></a> 的长文
<strong>RL Interview Questions 2026</strong> (原文链接) 与 <a
href="https://x.com/vivek_2332/status/2063566811749331353"><span
class="citation" data-cites="vivek_2332">@vivek_2332</span></a>
的逐题答案。</p>
<p>sheriyuo 在知乎也发过中文版: <a
href="https://zhuanlan.zhihu.com/p/2046740446353811230"
class="uri">https://zhuanlan.zhihu.com/p/2046740446353811230</a></p>
<p>sheriyuo 在原文中提到几点: 没严格区分 LLM RL 与 Agentic RL;
几乎每道题都能继续追问, 没有给出参考答案; 现代 RL
招聘越来越要求全栈理解, 算法岗也会问 infra, 反之亦然; 数据相关问题没收录
(太看经验); 死记硬背不够, 深度理解更重要。</p>]]></description>
</item>
</channel>
</rss>
