<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>MAGMA -- Multimodal Augmentation of Generative Models through Adapter-based Finetuning · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2112.05253/</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/2112.05253/feed.xml" rel="self" type="application/rss+xml" /><item><title>MAGMA -- Multimodal Augmentation of Generative Models through Adapter-based Finetuning — HN submission</title><link>https://news.ycombinator.com/item?id=33956930</link><guid isPermaLink="false">urn:hackerpapers:submission:2112.05253:33956930</guid><pubDate>Mon, 12 Dec 2022 16:51:41 GMT</pubDate><description>&lt;p&gt;Ask HN: Running (multimodal) semantic search on edge?&lt;/p&gt;&lt;p&gt;Has anyone ever tried to run (multimodal) semantic search in ONNX? (or alternative, i.e. mobile/JS) Do you think multimodal models will increase in size over time? (even less likely to run on edge?) For example, https://arxiv.org/abs/2112.05253 seems to go in the direction of bigger models Why language seems to imply large models but image models seem usually smaller?&lt;/p&gt;&lt;p&gt;1 point; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2112.05253/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.CV</category></item></channel></rss>
