<rss version="2.0">
  <channel>
    <title>r-coding_agents on superterran.net</title>
    <link>https://superterran.net/categories/r-coding_agents/</link>
    <description></description>
    
    <language>en</language>
    
    <lastBuildDate>Mon, 11 May 2026 19:54:34 -0400</lastBuildDate>
    
    <item>
      <title></title>
      <link>https://superterran.net/2026/05/11/acc9d8.html</link>
      <pubDate>Mon, 11 May 2026 19:54:34 -0400</pubDate>
      
      <guid>http://superterran.micro.blog/2026/05/11/acc9d8.html</guid>
      <description>&lt;p&gt;Languages have a ton of overlap, a language model trained on all repos would in principle be stronger than one trained on a small subset language.&lt;/p&gt;
&lt;p&gt;&lt;a href=&#34;https://www.reddit.com/r/coding_agents/comments/1t8zr2u/why_not_language_specific_slms_as_coding_agent/ol9zn6f/&#34;&gt;www.reddit.com/r/coding_&amp;hellip;&lt;/a&gt;&lt;/p&gt;
</description>
    </item>
    
  </channel>
</rss>