Simple feedback system for programming assignment to improve idiomatic use of code

The aim of this master’s thesis is to investigate the utilization of pre-trained large language models in the domain of programming education to enhance students’ programming skills, particularly focusing on the idiomatic use of code. We propose the development of an automated feedback generation system that harnesses the power of pre-trained language models, such as GPT-3 and starcoder, to provide personalized and actionable feedback to students on their code submissions. By analyzing the code syntactically, semantically, and through contextual understanding, the system will offer suggestions and recommendations to improve the idiomatic use of a code. To achieve this, the thesis explore and researches fitting models, the process of fine-tuning and adapting pre-trained language models to programming-specific contexts. furthermore compares results achieved by the selected model and suggests future work to improve results. The findings of this research will contribute to the field of programming education by demonstrating the potential of pre-trained language models in providing tailored and constructive feedback on code submissions. This thesis will shed light on the benefits and limitations of leveraging these models, paving the way for future advancements in automated feedback systems and promoting the development of more idiomatic programming skills among students.