<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>nGPT: Normalized Transformer with Representation Learning on the Hypersphere · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2410.01131/</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/2410.01131/feed.xml" rel="self" type="application/rss+xml" /><item><title>nGPT: Normalized Transformer with Representation Learning on the Hypersphere — mentioned by szcs</title><link>https://news.ycombinator.com/item?id=42018030</link><guid isPermaLink="false">urn:hackerpapers:mention:2410.01131:42018030</guid><pubDate>Fri, 01 Nov 2024 15:37:50 GMT</pubDate><description>&lt;p&gt;There is a particularly nice geometric interpretation of attention I just realised recently in a flash of enlightenment, best explained with an interactive Desmos plot (black dot is draggable): https://www.desmos.com/calculator/3rtqsyapxo The above assumes the columns of K are normalised but bear with me. K and V together form a vector database. V are the payloads, each row containing a vector of data. K describes the position of these points in space, on the surface of a hypershpere. The query vector describes the query into the database: the vector direction describes the point in space…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2410.01131/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.LG</category></item><item><title>nGPT: Normalized Transformer with Representation Learning on the Hypersphere — HN submission</title><link>https://news.ycombinator.com/item?id=41881478</link><guid isPermaLink="false">urn:hackerpapers:submission:2410.01131:41881478</guid><pubDate>Fri, 18 Oct 2024 17:18:33 GMT</pubDate><description>&lt;p&gt;NGPT: Normalized Transformer with Representation Learning on the Hypersphere&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;4 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2410.01131/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.LG</category></item></channel></rss>
