<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Composing Modeling and Inference Operations with Probabilistic Program Combinators · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1811.05965/</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/1811.05965/feed.xml" rel="self" type="application/rss+xml" /><item><title>Composing Modeling and Inference Operations with Probabilistic Program Combinators — mentioned by eli_gottlieb</title><link>https://news.ycombinator.com/item?id=20225166</link><guid isPermaLink="false">urn:hackerpapers:mention:1811.05965:20225166</guid><pubDate>Wed, 19 Jun 2019 17:17:44 GMT</pubDate><description>&lt;p&gt;&amp;gt;As a side note, I&amp;#x27;ve been doing more probabilistic programming with pymc3 recently, and it&amp;#x27;s pretty incredible how leaky the abstractions can be. I&amp;#x27;m not saying there&amp;#x27;s a way to do better, just that at present there&amp;#x27;s a huge gap between the beautiful vision of &amp;quot;the inference button&amp;quot; and the current tools. Yeah, there&amp;#x27;s a pretty active debate in the probabilistic programming R&amp;amp;D community over whether it&amp;#x27;s a bug to sell people on &amp;quot;the inference button&amp;quot;, then deliver a leaky abstraction, or a feature to offer richly programmable inference. Our lab has been working on some ideas to get &amp;quot;basic&amp;quot;…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1811.05965/"&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>
