| dc.rights.license | CC-BY-NC-ND | |
| dc.contributor.advisor | Jeuring, Johan | |
| dc.contributor.author | Wannee, Laurian | |
| dc.date.accessioned | 2024-03-15T00:01:33Z | |
| dc.date.available | 2024-03-15T00:01:33Z | |
| dc.date.issued | 2024 | |
| dc.identifier.uri | https://studenttheses.uu.nl/handle/20.500.12932/46163 | |
| dc.description.sponsorship | Utrecht University | |
| dc.language.iso | EN | |
| dc.subject | The work aims to answer the question: “Does specifying sub-goals in the prompts for GitHub Copilot and ChatGPT improve the quality of generated code for Python problems that should be solvable by students that have completed CS1?” Using a dataset of 166 Python programming problems. The models are first prompted using the standard problem description, and then using a sub-goal approach to measure how useful these are in improving the output of these models. | |
| dc.title | Enhancing natural-language prompts for code completion tools using sub goals | |
| dc.type.content | Master Thesis | |
| dc.rights.accessrights | Open Access | |
| dc.subject.keywords | Large Language Models, Sub-goals, Python, Prompt Engineering, GPT-3, Code Generation | |
| dc.subject.courseuu | Business Informatics | |
| dc.thesis.id | 29170 | |
