An Empirical Evaluation of Using Large Language Models for Automated Unit Test Generation