# Reward Model

> Source: https://sukruyusufkaya.com/en/glossary/reward-model
> Updated: 2026-05-13T20:58:57.237Z
> Type: glossary
> Category: uretken-yapay-zeka-ve-llm
**TLDR:** An auxiliary model that estimates how preferable generated outputs are and provides signals for alignment.

<p>A reward model is the learned evaluation mechanism used in RLHF-style alignment to estimate which output is better. It converts human preferences into a continuous training signal. However, if the reward model is biased or brittle, it can affect the entire alignment process. It is therefore a critical but sensitive component.</p>