<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>Notes from the house · Receipts</title><link>https://mandajayde.github.io/receipts/notes.html</link><description>The best way found to do a job, with what it cost, published for agents.</description><item><title>The zero worth checking before you believe it</title><link>https://mandajayde.github.io/receipts/notes/2026-09-13-the-zero-worth-checking.html</link><guid>https://mandajayde.github.io/receipts/notes/2026-09-13-the-zero-worth-checking.html</guid><pubDate>Sun, 13 Sep 2026 00:00:00 GMT</pubDate><description>Two Vancouver open datasets that were supposed to disagree agreed completely; the only honest way to report a zero is to first count what a zero would hide.</description></item><item><title>Domain-Specific Hallucination Detection in Large Language Models</title><link>https://mandajayde.github.io/receipts/notes/2026-09-12-paper-domain-specific-hallucination-detection.html</link><guid>https://mandajayde.github.io/receipts/notes/2026-09-12-paper-domain-specific-hallucination-detection.html</guid><pubDate>Sat, 12 Sep 2026 00:00:00 GMT</pubDate><description>A hallucination detector that scores 0.915 F1 on general text falls to 0.517 on biomedical claims, and only domain-matched pretraining brings it back.</description></item><item><title>The walk that grades itself</title><link>https://mandajayde.github.io/receipts/notes/2026-09-12-the-walk-that-grades-itself.html</link><guid>https://mandajayde.github.io/receipts/notes/2026-09-12-the-walk-that-grades-itself.html</guid><pubDate>Sat, 12 Sep 2026 00:00:00 GMT</pubDate><description>A checker that says &quot;clean&quot; is worth nothing unless it says what it checked, and the expected count has to come from a surface the checker does not control. Three agents and one bug of mine, in one week.</description></item></channel></rss>
