<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>Autonomous Research Agents on AIBussin — AI applications, systems and books</title>
		<link>https://aibussin.com/tags/autonomous-research-agents/</link>
		<description>Recent content in Autonomous Research Agents on AIBussin — AI applications, systems and books</description>
		<generator>Hugo</generator>
		<language>en-US</language>
		
		
		
		
			<lastBuildDate>Mon, 09 Jun 2025 10:00:03 +0100</lastBuildDate>
		
			<atom:link href="https://aibussin.com/tags/autonomous-research-agents/index.xml" rel="self" type="application/rss+xml" />
			<item>
				<title>Dimensions of Thought: A Smarter Way to Evaluate AI</title>
				<link>https://aibussin.com/post/dimensions/</link>
				<pubDate>Mon, 09 Jun 2025 10:00:03 +0100</pubDate>
				<guid>https://aibussin.com/post/dimensions/</guid>
				<description>&lt;h2 id=&#34;-summary&#34;&gt;📖 Summary&lt;/h2&gt;&#xA;&lt;p&gt;This post introduces a multidimensional reward modeling pipeline built on top of the &lt;a href=&#34;https://github.com/ernanhughes/co-ai&#34;&gt;stephanieanie&lt;/a&gt; framework. It covers:&lt;/p&gt;&#xA;&lt;ul&gt;&#xA;&lt;li&gt;&#xA;&lt;p&gt;✅ &lt;strong&gt;Structured Evaluation Setup&lt;/strong&gt;&#xA;How to define custom evaluation dimensions using YAML or database-backed rubrics.&lt;/p&gt;&#xA;&lt;/li&gt;&#xA;&lt;li&gt;&#xA;&lt;p&gt;🧠 &lt;strong&gt;Automated Scoring with LLMs&lt;/strong&gt;&#xA;Using the &lt;code&gt;ScoreEvaluator&lt;/code&gt; to produce structured, rationale-backed scores for each dimension.&lt;/p&gt;&#xA;&lt;/li&gt;&#xA;&lt;li&gt;&#xA;&lt;p&gt;🧮 &lt;strong&gt;Embedding-Based Hypothesis Indexing&lt;/strong&gt;&#xA;Efficiently embedding hypotheses and comparing them for contrastive learning using similarity.&lt;/p&gt;&#xA;&lt;/li&gt;&#xA;&lt;li&gt;&#xA;&lt;p&gt;🔄 &lt;strong&gt;Contrast Pair Generation&lt;/strong&gt;&#xA;Creating training pairs where one hypothesis outperforms another on a given dimension.&lt;/p&gt;</description>
			</item>
	</channel>
</rss>
