<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2407.14494/</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/2407.14494/feed.xml" rel="self" type="application/rss+xml" /><item><title>InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques — mentioned by ag8</title><link>https://news.ycombinator.com/item?id=50067898</link><guid isPermaLink="false">urn:hackerpapers:mention:2407.14494:50067898</guid><pubDate>Tue, 13 Oct 2026 07:11:32 GMT</pubDate><description>&lt;p&gt;This is true in the immediate case, but unfortunately there are relatively simple methods to obfuscate weight distributions; see &amp;quot;Obfuscating compiled weights&amp;quot; on page 13, or Figure 2 here: https://arxiv.org/pdf/2407.14494&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2407.14494/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.LG</category></item></channel></rss>
