<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Federated Learning of Gboard Language Models with Differential Privacy · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2305.18465/</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/2305.18465/feed.xml" rel="self" type="application/rss+xml" /><item><title>Federated Learning of Gboard Language Models with Differential Privacy — mentioned by yujonglee</title><link>https://news.ycombinator.com/item?id=44727616</link><guid isPermaLink="false">urn:hackerpapers:mention:2305.18465:44727616</guid><pubDate>Tue, 29 Jul 2025 19:59:31 GMT</pubDate><description>&lt;p&gt;&amp;gt; How does it choose what&amp;#x27;s important vs what isn&amp;#x27;t? The idea of Hyprnote is that you write chicken-scratch-raw note during the meeting(what you think is important), and AI enhance based on it. On-device learning is interesting too. For example, Gboard: https://arxiv.org/abs/2305.18465 And yes - we are open to this too&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2305.18465/"&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>
