Comparative Studies: Leveraging Large Language Model In Theoritical and Practical Assessment Sample Question-Answer Bank on Programming Related Subjects

Practical Code Assessment has been important in assessing students' level of understanding, coding, and assessing/evaluating. One of the challenges faced by lecturers is the difficulty in producing questions and answers creatively. To address this issue, an experimental proof-of-concept (POC) prototype was developed using Natural Language Processing (NLP) and Prompt Engineering techniques in the Large Language Model (LLM) to generate practical and high-quality questions and answers. Two Large Language Models, namely LLaMa2-7b and Mixtral-7b were compared to provide a scalable POC solution. Through evaluation and benchmarking, the prototype was tested based on Human-level Performance (HLP) standards. This prototype enhanced the work efficiency of the lecturers as it shortened the time to create questions and answers for assessments, which eventually enhanced learning outcomes and the educational, learning, and teaching experience.