This paper provides a comprehensive exploration of evaluation metrics for Large Language Models (LLMs), focusing on their mathematical formulations and statistical interpretations. The authors categorize these metrics into three types: Multiple-Classification (MC), Token-Similarity (TS), and Question-Answering (QA) metrics. They provide detailed mathematical formulations and statistical explanations for each type of metric, highlighting their strengths and weaknesses. The paper also discusses the application of these metrics in recent biomedical LLMs, using case studies to illustrate their practical use. Additionally, the authors identify repositories containing the discussed metrics and offer a comparison of these metrics to aid researchers in selecting appropriate metrics for diverse tasks. The main contributions of the paper include summarizing and categorizing the metrics, providing mathematical formulations and statistical interpretations, identifying repositories, and showcasing the application of these metrics in biomedical LLMs. The paper concludes with a discussion of the challenges and limitations of current evaluation methods, emphasizing the need for more comprehensive and robust metrics to address issues such as imperfect gold standards and the lack of statistical inference methods.This paper provides a comprehensive exploration of evaluation metrics for Large Language Models (LLMs), focusing on their mathematical formulations and statistical interpretations. The authors categorize these metrics into three types: Multiple-Classification (MC), Token-Similarity (TS), and Question-Answering (QA) metrics. They provide detailed mathematical formulations and statistical explanations for each type of metric, highlighting their strengths and weaknesses. The paper also discusses the application of these metrics in recent biomedical LLMs, using case studies to illustrate their practical use. Additionally, the authors identify repositories containing the discussed metrics and offer a comparison of these metrics to aid researchers in selecting appropriate metrics for diverse tasks. The main contributions of the paper include summarizing and categorizing the metrics, providing mathematical formulations and statistical interpretations, identifying repositories, and showcasing the application of these metrics in biomedical LLMs. The paper concludes with a discussion of the challenges and limitations of current evaluation methods, emphasizing the need for more comprehensive and robust metrics to address issues such as imperfect gold standards and the lack of statistical inference methods.