<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Tiny Gradients</title><link>https://tinygradients.com/</link><description>Recent content on Tiny Gradients</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Tue, 29 Sep 2026 10:30:00 -0400</lastBuildDate><atom:link href="https://tinygradients.com/index.xml" rel="self" type="application/rss+xml"/><item><title>A study roadmap for LLMs and agents, ordered by what depends on what</title><link>https://tinygradients.com/posts/study-roadmap/</link><pubDate>Tue, 29 Sep 2026 10:30:00 -0400</pubDate><guid>https://tinygradients.com/posts/study-roadmap/</guid><description>Reading lists are ordered by source. This one is ordered by what each idea needs first, from autodiff to agent security.</description></item><item><title>Hello, gradients</title><link>https://tinygradients.com/posts/hello-gradients/</link><pubDate>Tue, 29 Sep 2026 09:00:00 -0400</pubDate><guid>https://tinygradients.com/posts/hello-gradients/</guid><description>What this notebook is for, and why the steps are small.</description></item><item><title>About</title><link>https://tinygradients.com/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://tinygradients.com/about/</guid><description>&lt;p&gt;Tiny Gradients is a notebook about learning AI from the ground up: the basics of&#10;neural networks, then language models, then agents and how to evaluate them.&lt;/p&gt;&#10;&lt;p&gt;I write it while I study, so expect small steps, corrections, and the occasional&#10;wrong turn. The name comes from gradient descent, where progress is a long series&#10;of small, careful steps, each one a little closer than the last.&lt;/p&gt;&#10;&lt;p&gt;To get new posts, subscribe to the &lt;a href="https://tinygradients.com/index.xml"&gt;RSS feed&lt;/a&gt;.&lt;/p&gt;</description></item></channel></rss>