<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Posts on Matěj Kripner</title><link>https://kripner.com/posts/</link><description>Recent content in Posts on Matěj Kripner</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 01 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://kripner.com/posts/index.xml" rel="self" type="application/rss+xml"/><item><title>Playing Chess with an SO-101 Robot Arm</title><link>https://kripner.com/posts/chess-robot/</link><pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/chess-robot/</guid><description>Teaching π0.5 to play physical chess with the SO-101 robot arm: How I got to a working demo and what is planned next.</description></item><item><title>Papers Week</title><link>https://kripner.com/posts/papers-week/</link><pubDate>Tue, 17 Feb 2026 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/papers-week/</guid><description>Reading five papers in five days: Kimi K2.5, DeepMind&amp;rsquo;s work on AI4MATH, QED-Nano, Numina-Lean-Agent, and I-JEPA.</description></item><item><title>Let's see how TRM reasons</title><link>https://kripner.com/posts/trm-rollouts/</link><pubDate>Fri, 31 Oct 2025 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/trm-rollouts/</guid><description>Looking inside the Tiny Recursive Model&amp;rsquo;s latent recursion traces on ARC, with the hypothesis that its intermediate states decompose a task into reusable concepts.</description></item><item><title>Paper notes: σ-GPTs, A New Approach to Autoregressive Models</title><link>https://kripner.com/posts/sigma-gpt/</link><pubDate>Sat, 14 Sep 2024 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/sigma-gpt/</guid><description>σ-GPTs generate a sequence in any order, not just left to right — two small changes to a standard GPT, and a rejection-sampling trick that fills in many tokens per pass.</description></item><item><title>Paper notes: Longformer, The Long-Document Transformer</title><link>https://kripner.com/posts/longformer/</link><pubDate>Thu, 23 Nov 2023 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/longformer/</guid><description>Longformer replaces full self-attention with a sliding window plus a few global tokens, cutting the quadratic cost down to linear in sequence length.</description></item><item><title>Paper notes: Reasoning over Paths in Knowledge Bases using Reinforcement Learning</title><link>https://kripner.com/posts/minerva/</link><pubDate>Tue, 07 Nov 2023 00:00:00 +0000</pubDate><guid>https://kripner.com/posts/minerva/</guid><description>MINERVA answers knowledge-graph queries by learning to walk the graph: a policy-gradient agent that picks one outgoing edge at a time instead of enumerating every possible path.</description></item></channel></rss>