University of Twente Student Theses


Fine-tuning GPT-3 for data visualization code generation

Bosman, Joris (2023) Fine-tuning GPT-3 for data visualization code generation.

[img] PDF
Abstract:Data and data visualization play crucial roles in the field of technology. However, data visualization is not always a simple and quick process, and it often involves a learning curve. To address this challenge and enhance the efficiency of creating graphical displays, one approach is to utilize a GPT (Generative Pre-trained Transformer). GPTs are trained to understand natural language prompts and generate responses, including code snippets. By providing specific instructions in natural language, the GPT can generate code tailored to a data visualization library, which can then interpret and create the desired visual representation of the data. It's important to note that while GPTs are powerful tools, they are not without flaws and may make mistakes. However, by fine-tuning a GPT, we can improve its performance and reduce the occurrence of errors. Therefore, this project aims to investigate the effectiveness of fine-tuning GPT-3 specifically for data visualization tasks. By evaluating its capabilities and limitations, we can gain insights into the potential of leveraging GPTs for more efficient and accurate data visualization processes.
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:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page