<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Human Feedback on AI VOID</title><link>https://ai-blog.noorshomelab.dev/tags/human-feedback/</link><description>Recent content in Human Feedback on AI VOID</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 30 Jan 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://ai-blog.noorshomelab.dev/tags/human-feedback/index.xml" rel="self" type="application/rss+xml"/><item><title>Chapter 7: Introduction to Reinforcement Learning from Human Feedback (RLHF) Concepts</title><link>https://ai-blog.noorshomelab.dev/tunix-mastery-2026/07-rlhf-concepts/</link><pubDate>Fri, 30 Jan 2026 00:00:00 +0000</pubDate><guid>https://ai-blog.noorshomelab.dev/tunix-mastery-2026/07-rlhf-concepts/</guid><description>&lt;h2 id="introduction-to-reinforcement-learning-from-human-feedback-rlhf-concepts"&gt;Introduction to Reinforcement Learning from Human Feedback (RLHF) Concepts&lt;/h2&gt;
&lt;p&gt;Welcome to Chapter 7! So far, we&amp;rsquo;ve explored the foundational aspects of Tunix, understanding how it leverages JAX to efficiently manage and fine-tune Large Language Models (LLMs). We&amp;rsquo;ve touched upon pre-training and various forms of supervised fine-tuning. But what happens when you want your LLM to not just generate coherent text, but to also be &lt;em&gt;helpful&lt;/em&gt;, &lt;em&gt;harmless&lt;/em&gt;, and &lt;em&gt;honest&lt;/em&gt;—to truly align with human values and instructions? That&amp;rsquo;s where Reinforcement Learning from Human Feedback, or RLHF, steps in.&lt;/p&gt;</description></item></channel></rss>