LLM-as-Judge

Definitions

1. An evaluation pattern where one language model grades, compares, or scores the outputs of another model according to a rubric or prompt.

   In plain English: Using one AI model to evaluate another AI model’s answers.

   Example: "The team used LLM-as-judge for fast pairwise ranking before human review."

Related Terms

• Benchmark Contamination
• AI Benchmark
• AI Fairness
• AI Hallucination
• AI Model Card
• Anomaly Detection ML
• Context Window Limit
• DeepSeek
• Retrieval Augmented Generation
• LLM Benchmark
• LLM Judge