<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <link type="text/html" rel="alternate" href="http://ln.ht/substack/article"/>
  <link type="application/atom+xml" rel="self" href="http://ln.ht/_/feed/substack/article"/>
  <id>http://ln.ht/_/feed/substack/article</id>
  <title>Bookmarks tagged with: substack,article</title>
  <updated>2026-06-29T11:46:52.613677Z</updated>
  <entry>
    <category label="substack" term="substack"/>
    <category label="article" term="article"/>
    <category label="attention" term="attention"/>
    <author>
      <name>cos</name>
      <uri>https://ln.ht/~cos</uri>
    </author>
    <content></content>
    <link rel="alternate" href="https://www.henrikkarlsson.xyz/p/attention"/>
    <id>https://www.henrikkarlsson.xyz/p/attention</id>
    <title>Almost anything you give sustained attention to will begin to loop on itself and bloom</title>
    <updated>2025-09-04T18:40:53Z</updated>
  </entry>
  <entry>
    <category label="llm" term="llm"/>
    <category label="article" term="article"/>
    <category label="substack" term="substack"/>
    <author>
      <name>edwardloveall</name>
      <uri>https://ln.ht/~edwardloveall</uri>
    </author>
    <content>On a dataset of programming challenges, GPT-4 solved 10 out of 10 problems that had been published before 2021 (GPT-4’s pre-training cutoff date) and zero out of 10 problems that had been published after 2021.</content>
    <link rel="alternate" href="https://aiguide.substack.com/p/can-large-language-models-reason"/>
    <id>https://aiguide.substack.com/p/can-large-language-models-reason</id>
    <title>GPT-4 fails all tests that came out after it was trained</title>
    <updated>2023-09-18T21:24:05Z</updated>
  </entry>
</feed>