<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/c404695b4b5d4267be2eada7c91e40b3&quot; frameborder=&quot;0&quot; width=&quot;1280&quot; height=&quot;960&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>960</height><width>1280</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>960</thumbnail_height><thumbnail_width>1280</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/c404695b4b5d4267be2eada7c91e40b3-70b2a3f5f44c93ca.gif</thumbnail_url><duration>281.223</duration><title>Local RAC Pipeline for Legal PDFs</title><description>Hi, I am Aryan Sana, and this short demo walks through section 2, a local RAC pipeline for querying legal PDF documents. I run an evaluation harness that checks retrieval quality using precision at 3, and on the sample data the score is 1, with all 10 queries retrieving the correct source within the top 3. Then I demo a question about the notice period, showing a confidence value of 0.66 with snippets and cited pages. I also test a non related query and show low confidence with no sufficient evidence found. I do not request any action from viewers.</description></oembed>