<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Tokenization on Anton Chirikalov's Blog</title><link>https://antonch.me/en/tags/tokenization/</link><description>Recent content in Tokenization on Anton Chirikalov's Blog</description><generator>Hugo -- 0.164.0</generator><language>en-US</language><lastBuildDate>Fri, 10 Jul 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://antonch.me/en/tags/tokenization/index.xml" rel="self" type="application/rss+xml"/><item><title>Dissecting the Transformer. Part 1</title><link>https://antonch.me/en/posts/transformer-anatomy-part-1/</link><pubDate>Fri, 10 Jul 2026 00:00:00 +0000</pubDate><guid>https://antonch.me/en/posts/transformer-anatomy-part-1/</guid><description>A technical walkthrough: how the tokenizer slices a prompt, how a token becomes a vector, why position is encoded as rotation (RoPE), and how attention works — where the Q, K, V matrices come from and what they do</description></item></channel></rss>