<?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>Probing on Cognaptus</title>
    <link>https://cognaptus.com/tags/probing/</link>
    <description>Recent content in Probing on Cognaptus</description>
    <generator>Hugo -- 0.145.0</generator>
    <language>en-us</language>
    <lastBuildDate>Fri, 02 Oct 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://cognaptus.com/tags/probing/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Readable Before Steerable: Why Probe Accuracy Is Not a Control Test</title>
      <link>https://cognaptus.com/blog/2026-10-02-readable-before-steerable-why-probe-accuracy-is-not-a-control-test/</link>
      <pubDate>Fri, 02 Oct 2026 00:00:00 +0000</pubDate>
      <guid>https://cognaptus.com/blog/2026-10-02-readable-before-steerable-why-probe-accuracy-is-not-a-control-test/</guid>
      <description>A developmental study of language models shows why highly accurate internal probes should not be treated as evidence that the same representation can reliably steer behavior.</description>
    </item>
  </channel>
</rss>
