<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>LM Judges on Programmer.ie</title>
    <link>http://programmer.ie/tags/lm-judges/</link>
    <description>Recent content in LM Judges on Programmer.ie</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Fri, 28 Aug 2026 10:40:00 +0100</lastBuildDate>
    <atom:link href="http://programmer.ie/tags/lm-judges/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>DSPy From First Principles 09: The Metric Is Part of the Program</title>
      <link>http://programmer.ie/books/dspy-from-first-principles/09-chapter/</link>
      <pubDate>Fri, 28 Aug 2026 10:40:00 +0100</pubDate>
      <guid>http://programmer.ie/books/dspy-from-first-principles/09-chapter/</guid>
      <description>&lt;h1 id=&#34;the-metric-is-part-of-the-program&#34;&gt;The Metric Is Part of the Program&lt;/h1&gt;&#xA;&lt;p&gt;Chapter 8 gave us a baseline harness. It deliberately used a weak smoke-test metric because the first job was to make evaluation run.&lt;/p&gt;&#xA;&lt;p&gt;Now the dangerous part begins.&lt;/p&gt;&#xA;&lt;p&gt;During optimization, DSPy searches for program states that score well under the metric we provide. The optimizer does not know whether the score is a faithful proxy for the behavior we actually want.&lt;/p&gt;&#xA;&lt;hr&gt;&#xA;&lt;h2 id=&#34;1-the-metric-ladder&#34;&gt;1. The metric ladder&lt;/h2&gt;&#xA;&lt;p&gt;Different tasks tolerate different metrics.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
