<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>MULTITuDE: Large-Scale Multilingual Machine-Generated Text Detection Benchmark · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2310.13606/</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/2310.13606/feed.xml" rel="self" type="application/rss+xml" /><item><title>MULTITuDE: Large-Scale Multilingual Machine-Generated Text Detection Benchmark — mentioned by berdario</title><link>https://news.ycombinator.com/item?id=43150474</link><guid isPermaLink="false">urn:hackerpapers:mention:2310.13606:43150474</guid><pubDate>Sun, 23 Feb 2025 16:25:17 GMT</pubDate><description>&lt;p&gt;&amp;gt; They give useful probabilities Yes, compared to an all-or-nothing approach, it&amp;#x27;s better to be upfront about the uncertainty, especially if the tool surfaces the probability by-sentence. But how are those probabilities computed? You mention gptzero, but https://gptzero.me/technology doesn&amp;#x27;t clarify at all how it works. They link papers using GPTZero (i.e. from other researchers), e.g. https://arxiv.org/pdf/2310.13606 And these very same papers highlight how everything is still unknown &amp;gt; Despite their wide use and support of non-English languages, the extent of their zero- shot multilingual…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2310.13606/"&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>
