<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Speculative Decoding on Victor&#39;s Blog</title>
    <link>https://chzhyang.github.com/tags/speculative-decoding/</link>
    <description>Recent content in Speculative Decoding on Victor&#39;s Blog</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>en</language>
    <copyright>© 2026 Victor Yang</copyright>
    <lastBuildDate>Sat, 01 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://chzhyang.github.com/tags/speculative-decoding/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Speculative Decoding: Medusa, MTP, EAGLE, DFlash, DDTree, and DSpark</title>
      <link>https://chzhyang.github.com/tech/llm/speculative-decoding-medusa-mtp-eagle-dflash-ddtree-dspark/</link>
      <pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate>
      
      <guid>https://chzhyang.github.com/tech/llm/speculative-decoding-medusa-mtp-eagle-dflash-ddtree-dspark/</guid>
      <description>Speculative decoding changes the unit of progress in LLM inference: a target-model forward pass can commit several tokens instead of one.</description>
      
    </item>
    
  </channel>
</rss>
