Fine-Tuning and Evaluating Open-Source Large Language Models for the Army Domain