<?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/53b63c0e808d42b8aca9fe614963817a&quot; frameborder=&quot;0&quot; width=&quot;1960&quot; height=&quot;1470&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1470</height><width>1960</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1470</thumbnail_height><thumbnail_width>1960</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/53b63c0e808d42b8aca9fe614963817a-1687771233848.gif</thumbnail_url><duration>264.4666666666666</duration><title>Few-shot prompting - Part 3: Publishing and deploying our app</title><description>Hi there! In this Loom, I&apos;m going to show you how to clean up the output of our app that uses FuseShop prompting to teach the LLM to respond in a way that we want it to respond. We&apos;ll start by adding a description to make our form look nicer and tighter. Then, we&apos;ll configure our output to have a single output which is the answer, and we&apos;ll select a variable to be the answer from the LM step previously. This will reduce all the steps that are coming out here into a single one called answer, and it&apos;s going to be linking to the variable answer from the LM step. Once we&apos;ve done that, we&apos;ll go back to the use page and re-trigger this. No action is requested from you, but this will help you understand how to clean up the output of our app.</description></oembed>