This paper provides a comprehensive overview of metrics used to evaluate Large Language Models (LLMs), focusing on their mathematical formulations, statistical interpretations, and practical applications. The authors categorize metrics into three types: Multiple-Classification (MC), Token-Similarity (TS), and Question-Answering (QA). MC metrics assess the ability of LLMs to classify texts into multiple groups, while TS metrics evaluate the similarity between generated and reference texts. QA metrics are specifically designed for question-answering tasks. The paper discusses the strengths and limitations of each metric, highlighting issues such as imperfect gold standards and the lack of statistical inference methods. It also provides a detailed comparison of these metrics, along with their statistical interpretations and applications in biomedical LLMs. The authors emphasize the importance of selecting appropriate metrics for different tasks and highlight the need for more comprehensive evaluation methods that consider a broader range of metrics. The paper concludes with a call for further research to address the challenges in evaluating LLMs and to develop more reliable metrics for assessing their performance.This paper provides a comprehensive overview of metrics used to evaluate Large Language Models (LLMs), focusing on their mathematical formulations, statistical interpretations, and practical applications. The authors categorize metrics into three types: Multiple-Classification (MC), Token-Similarity (TS), and Question-Answering (QA). MC metrics assess the ability of LLMs to classify texts into multiple groups, while TS metrics evaluate the similarity between generated and reference texts. QA metrics are specifically designed for question-answering tasks. The paper discusses the strengths and limitations of each metric, highlighting issues such as imperfect gold standards and the lack of statistical inference methods. It also provides a detailed comparison of these metrics, along with their statistical interpretations and applications in biomedical LLMs. The authors emphasize the importance of selecting appropriate metrics for different tasks and highlight the need for more comprehensive evaluation methods that consider a broader range of metrics. The paper concludes with a call for further research to address the challenges in evaluating LLMs and to develop more reliable metrics for assessing their performance.