<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Searching for Activation Functions · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1710.05941/</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/1710.05941/feed.xml" rel="self" type="application/rss+xml" /><item><title>Searching for Activation Functions — mentioned by f_devd</title><link>https://news.ycombinator.com/item?id=34713359</link><guid isPermaLink="false">urn:hackerpapers:mention:1710.05941:34713359</guid><pubDate>Wed, 08 Feb 2023 19:21:12 GMT</pubDate><description>&lt;p&gt;Learnable activation functions are a thing famously Swish[0] is is a trainable SiLU which was found through symbolic search/optimization [1], but as it turns out that doesn&amp;#x27;t magically make make neural networks orders better. [0]: https://en.m.wikipedia.org/wiki/Swish_function [1]: https://arxiv.org/abs/1710.05941&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item><item><title>Searching for Activation Functions — mentioned by Hawkenfall</title><link>https://news.ycombinator.com/item?id=21545872</link><guid isPermaLink="false">urn:hackerpapers:mention:1710.05941:21545872</guid><pubDate>Fri, 15 Nov 2019 15:59:56 GMT</pubDate><description>&lt;p&gt;A more in-depth paper about this found the Swish activation often outperformed other functions: https://arxiv.org/abs/1710.05941&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item><item><title>Searching for Activation Functions — HN submission</title><link>https://news.ycombinator.com/item?id=15610146</link><guid isPermaLink="false">urn:hackerpapers:submission:1710.05941:15610146</guid><pubDate>Thu, 02 Nov 2017 13:22:47 GMT</pubDate><description>&lt;p&gt;The “Swish” Activation Function: Like ReLU but Better&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;2 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item><item><title>Searching for Activation Functions — HN submission</title><link>https://news.ycombinator.com/item?id=15587130</link><guid isPermaLink="false">urn:hackerpapers:submission:1710.05941:15587130</guid><pubDate>Mon, 30 Oct 2017 17:40:26 GMT</pubDate><description>&lt;p&gt;Swish: A Self-Gated Activation Function&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;2 points; 1 discussion reply on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item><item><title>Searching for Activation Functions — HN submission</title><link>https://news.ycombinator.com/item?id=15566846</link><guid isPermaLink="false">urn:hackerpapers:submission:1710.05941:15566846</guid><pubDate>Fri, 27 Oct 2017 09:24:24 GMT</pubDate><description>&lt;p&gt;Swish: A Self-Gated Activation Function&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;2 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item><item><title>Searching for Activation Functions — HN submission</title><link>https://news.ycombinator.com/item?id=15498001</link><guid isPermaLink="false">urn:hackerpapers:submission:1710.05941:15498001</guid><pubDate>Wed, 18 Oct 2017 10:19:53 GMT</pubDate><description>&lt;p&gt;Swish: A Self-Gated Activation Function&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;6 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/1710.05941/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.NE</category></item></channel></rss>
