Glossary · Methodology

LLM-as-Judge

An evaluation technique that uses one language model to score or grade the output of another, often used to scale evaluation beyond what manual human review can cover, with known limitations around judge-model bias.