<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>David Samuel — Blog</title>
  <link href="https://davidsamuel.no/feed.xml" rel="self"/>
  <link href="https://davidsamuel.no/"/>
  <id>https://davidsamuel.no/</id>
  <updated>2026-08-14T00:00:00Z</updated>
  <author>
    <name>David Samuel</name>
  </author>

  <entry>
    <title>Stop using greedy decoding for LLM inference</title>
    <link href="https://davidsamuel.no/blog/greedy-decoding/"/>
    <id>https://davidsamuel.no/blog/greedy-decoding/</id>
    <published>2026-08-14T00:00:00Z</published>
    <updated>2026-08-14T00:00:00Z</updated>
    <summary>Greedy decoding is not deterministic in practice, it is heavily biased, and its target is degenerate – why temperature=0 is a poor default for LLM inference.</summary>
  </entry>
</feed>
