<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · GVS5H</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/gvs5h/</link>
  <description>Material changes in AI models, agents, engineering and infrastructure, with primary sources and recommendations.</description>
  <language>en</language>
  <lastBuildDate>Mon, 02 Nov 2026 16:56:00 GMT</lastBuildDate>
  <atom:link href="https://derekwang2002.github.io/ai-intelligence/en/feeds/projects/gvs5h.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>GVS5H opens a ledger-based harness for repeated coding-agent passes</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20260826-06/</link>
    <guid isPermaLink="false">ev-20260826-06-published</guid>
    <pubDate>Sun, 25 Oct 2026 15:57:07 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;GVS5H runs a coding agent for five passes and uses a shared ledger for hypotheses, failures and next steps instead of adding a coordinator model. Code and run data are public. On 100 hard LiveCodeBench tasks, the authors report Qwen3.8-27B improving from 66.8 to 92.4.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Why read: &lt;/strong&gt;A shared failure ledger makes multi-pass agent retries reproducible without simply adding another coordinator.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://github.com/slee-persis/GVS5H&quot;&gt;slee-persis/GVS5H&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
