University of Twente Student Theses
Optimizing Code Generation Models Efficiency Through Hyperparameter Tuning
Hoenink, Quirijn (2025) Optimizing Code Generation Models Efficiency Through Hyperparameter Tuning.
|
PDF
10MB |
| Abstract: | This study explores how hyperparameter settings influence the efficiency of large language models (LLMs), focusing on the trade-off between accuracy and power consumption in small- and medium-sized models. Using different hyperparameter configurations, we explore how temperature settings, top p values, and token limits influence model performance and energy efficiency. The study incorporated 12 models from four different families, analyzing the correlation between these settings and their impact on accuracy and energy usage. The results indicate that selecting specific configurations can create more energy-efficient models without compromising performance. Further analysis will examine additional models and configurations to identify further optimization opportunities. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/105077 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page
