<?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>Genetic Improvement on Giovanni Pinna</title>
    <link>https://giovannipinna.net/tags/genetic-improvement/</link>
    <description>Recent content in Genetic Improvement on Giovanni Pinna</description>
    <generator>Hugo</generator>
    <language>en</language>
    <lastBuildDate>Tue, 01 Jul 2025 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://giovannipinna.net/tags/genetic-improvement/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Making the LLM-Plus-Evolution Pipeline Actually Smart</title>
      <link>https://giovannipinna.net/posts/sncs2025-exploring-gi-effect/</link>
      <pubDate>Tue, 01 Jul 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/sncs2025-exploring-gi-effect/</guid>
      <description>Last year we showed evolution can fix LLM code. This year we made the evolution itself smarter — better selection, partial credit, fewer cycles — and got improvements in 11 of 12 cases.</description>
    </item>
    <item>
      <title>Improving LLM-Generated Code via Genetic Improvement: A Summary of Recent Advances</title>
      <link>https://giovannipinna.net/posts/italia2025-gi-summary/</link>
      <pubDate>Mon, 23 Jun 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/italia2025-gi-summary/</guid>
      <description>A comprehensive summary of our research program on applying Genetic Improvement to LLM-generated code, presented at the Italian national AI conference Ital-IA 2025.</description>
    </item>
    <item>
      <title>What If We Stopped Asking ChatGPT to Fix Its Own Code?</title>
      <link>https://giovannipinna.net/posts/eurogp2024-gi-for-llm-code/</link>
      <pubDate>Wed, 03 Apr 2024 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/eurogp2024-gi-for-llm-code/</guid>
      <description>Self-correction is the default fix for buggy LLM code, but it has a ceiling. We tried something stranger — evolving the code instead — and it worked across every model we tested.</description>
    </item>
  </channel>
</rss>
