<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Reasoning on davethehuman.com</title><link>https://davethehuman.com/tags/reasoning/</link><description>Recent content in Reasoning on davethehuman.com</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Dave the human</copyright><lastBuildDate>Thu, 14 May 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://davethehuman.com/tags/reasoning/index.xml" rel="self" type="application/rss+xml"/><item><title>distillation</title><link>https://davethehuman.com/notes/distillation/</link><pubDate>Thu, 14 May 2026 00:00:00 +0000</pubDate><guid>https://davethehuman.com/notes/distillation/</guid><description>&lt;p&gt;&lt;em&gt;Distillation&lt;/em&gt; (also called &lt;a href="https://davethehuman.com/coming-soon/" rel="nofollow"&gt;knowledge distillation&lt;/a&gt;) consists of transferring complex reasoning patterns learned by larger models into smaller ones.
In &lt;a href="https://davethehuman.com/coming-soon/" rel="nofollow"&gt;deep learning&lt;/a&gt;, distillation happens when a smaller &amp;ldquo;student&amp;rdquo; model learns from outputs and logits of a larger &amp;ldquo;teacher&amp;rdquo; model; when talking about &lt;a href="https://davethehuman.com/notes/llm/"&gt;Large Language Models&lt;/a&gt;, distillation typically means performing &lt;a href="https://davethehuman.com/notes/supervised-fine-tuning/"&gt;supervised fine-tuning&lt;/a&gt; using high-quality labeled instruction datasets generated by a more capable &lt;a href="https://davethehuman.com/notes/llm/"&gt;LLM&lt;/a&gt;.&lt;/p&gt;</description></item></channel></rss>