LLM Judges

An evaluation method that uses large language models to assess the quality of AI-generated content. It is suited to scenarios where correctness cannot be determined by rigid rules, such as evaluating the stylistic quality of agent outputs or the appropriateness of tool-call decisions.