<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>RigidFormer: Learning Rigid Dynamics using Transformers · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2605.09196/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2605.09196/feed.xml" rel="self" type="application/rss+xml" /><item><title>RigidFormer: Learning Rigid Dynamics using Transformers — HN submission</title><link>https://news.ycombinator.com/item?id=48203485</link><guid isPermaLink="false">urn:hackerpapers:submission:2605.09196:48203485</guid><pubDate>Wed, 20 May 2026 05:33:08 GMT</pubDate><description>&lt;p&gt;RigidFormer: Learning Rigid Dynamics Using Transformers&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;1 point; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2605.09196/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.CV</category></item><item><title>RigidFormer: Learning Rigid Dynamics using Transformers — mentioned by E-Reverance</title><link>https://news.ycombinator.com/item?id=48199117</link><guid isPermaLink="false">urn:hackerpapers:mention:2605.09196:48199117</guid><pubDate>Tue, 19 May 2026 20:30:36 GMT</pubDate><description>&lt;p&gt;&amp;gt; But I am a bit reasurred that at least my job won&amp;#x27;t be fully replaced with AI :) I honestly can&amp;#x27;t comment with certainty that training from videos alone and whatever tokenization scheme they&amp;#x27;re using will ever get perfect dynamics. However it is worth noting that transformers can do a pretty good job at learning dynamics with the right pipeline (not video): https://arxiv.org/pdf/2605.15305 https://arxiv.org/pdf/2605.09196 My point here being that representationally, it might be possible to learn good dynamics without a radically different approach/arch. There are already models that extract…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2605.09196/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CV</category></item></channel></rss>
