<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AttentionViz: A Global View of Transformer Attention · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2305.03210/</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/2305.03210/feed.xml" rel="self" type="application/rss+xml" /><item><title>AttentionViz: A Global View of Transformer Attention — mentioned by kordlessagain</title><link>https://news.ycombinator.com/item?id=35930778</link><guid isPermaLink="false">urn:hackerpapers:mention:2305.03210:35930778</guid><pubDate>Sat, 13 May 2023 17:47:02 GMT</pubDate><description>&lt;p&gt;I&amp;#x27;ve been working on a project I&amp;#x27;ve called DocGPT: https://github.com/featurebasedb/DocGPT. This is a work in progress and I&amp;#x27;ll be putting together a release next week, once I&amp;#x27;ve got the instructions for running it done. It was my assumption that GPT-4 could be used for keyterm and question extraction from the fragments of a document and perhaps used to build a semantic graph that could be leveraged for rapid prompt assembly, augmenting the vector searches that are normally done. The idea here is to set &amp;quot;attention flags&amp;quot; on the document during discussion. In this example, I indexed the arxiv…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2305.03210/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.HC</category></item><item><title>AttentionViz: A Global View of Transformer Attention — mentioned by djoldman</title><link>https://news.ycombinator.com/item?id=35874637</link><guid isPermaLink="false">urn:hackerpapers:mention:2305.03210:35874637</guid><pubDate>Tue, 09 May 2023 14:14:37 GMT</pubDate><description>&lt;p&gt;About button: Attention Viz is an interactive tool that visualizes global attention patterns for transformer models. To create this tool, we visualize the joint embeddings of query and key vectors. paper: https://arxiv.org/abs/2305.03210 docs: https://catherinesyeh.github.io/attn-docs/&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2305.03210/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.HC</category></item><item><title>AttentionViz: A Global View of Transformer Attention — HN submission</title><link>https://news.ycombinator.com/item?id=35857512</link><guid isPermaLink="false">urn:hackerpapers:submission:2305.03210:35857512</guid><pubDate>Mon, 08 May 2023 02:40:05 GMT</pubDate><description>&lt;p&gt;AttentionViz: A Global View of Transformer Attention&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;3 points; 2 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2305.03210/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.HC</category></item></channel></rss>
