Metrics are a cornerstone element in evaluating any AI system, and in the case of large language models (LLMs), this is no exception.