<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Energy-Based Models on Nalar</title>
    <link>https://nalar.dev/tags/energy-based-models/</link>
    <description>Recent content in Energy-Based Models on Nalar</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Fri, 11 Sep 2026 00:00:00 +0700</lastBuildDate>
    <atom:link href="https://nalar.dev/tags/energy-based-models/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Score Candidates with Energy-Based Models</title>
      <link>https://nalar.dev/score-candidates-with-energy-based-models/</link>
      <pubDate>Fri, 11 Sep 2026 00:00:00 +0700</pubDate>
      <guid>https://nalar.dev/score-candidates-with-energy-based-models/</guid>
      <description>&lt;h1 id=&#34;score-candidates-with-energy-based-models&#34;&gt;Score Candidates with Energy-Based Models&lt;/h1&gt;&#xA;&lt;p&gt;Many AI systems need to decide which candidate fits an input: which reply matches a conversation, which label fits an image, or which configuration is plausible. A common design makes the model output a probability directly. &lt;strong&gt;Energy-based models&lt;/strong&gt; take a more general route: they assign each input-candidate pair a scalar energy, with lower values representing greater compatibility.&lt;/p&gt;&#xA;&lt;p&gt;That simple change is useful because the model can focus on relative preference without requiring every architecture to produce a normalized probability during scoring. It also introduces real engineering challenges. Training needs informative alternatives, probability normalization can be expensive, and inference may require searching a large candidate space.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
