<?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>Verifiable Rewards on Cognaptus</title>
    <link>https://cognaptus.com/tags/verifiable-rewards/</link>
    <description>Recent content in Verifiable Rewards on Cognaptus</description>
    <generator>Hugo -- 0.145.0</generator>
    <language>en-us</language>
    <lastBuildDate>Thu, 05 Mar 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://cognaptus.com/tags/verifiable-rewards/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Bending the Beam, Not the Brain: What RL with Perfect Rewards Still Can’t Teach LLMs</title>
      <link>https://cognaptus.com/blog/2026-03-05-bending-the-beam-not-the-brain-what-rl-with-perfect-rewards-still-cant-teach-llms/</link>
      <pubDate>Thu, 05 Mar 2026 00:00:00 +0000</pubDate>
      <guid>https://cognaptus.com/blog/2026-03-05-bending-the-beam-not-the-brain-what-rl-with-perfect-rewards-still-cant-teach-llms/</guid>
      <description>BeamPERL shows that exact physics rewards can specialize compact LLMs, but they do not automatically produce transferable scientific reasoning.</description>
    </item>
  </channel>
</rss>
