<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Continuous diffusion for categorical data · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2211.15089/</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/2211.15089/feed.xml" rel="self" type="application/rss+xml" /><item><title>Continuous diffusion for categorical data — mentioned by kastnerkyle</title><link>https://news.ycombinator.com/item?id=35302904</link><guid isPermaLink="false">urn:hackerpapers:mention:2211.15089:35302904</guid><pubDate>Sat, 25 Mar 2023 14:22:18 GMT</pubDate><description>&lt;p&gt;Discrete diffusion with rewriting can work well. It feels loosely similar to backtracking, if you assume n_steps large enough - need to be able to rewrite any non-provided position though I think (not all setups do this). Downside is the noise in discrete diffusion (in simplest case randomizing over all vocabulary space) is pretty harsh and makes things very difficult practically. Don&amp;#x27;t have an exact reference on the relationship, but feels similar to backtracking type mechanics in my experience. I found things tend to &amp;quot;lock in&amp;quot; quickly once a good path is found, which feels a lot like…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2211.15089/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CL</category></item></channel></rss>
