Mid Sweden University

miun.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Leveraging GPT-SW3 for Ethical and Legal AIApplications in Swedish Public Organizations: Training Methods and Challenges in Adapting GPT-SW3 to Local Needs
Mid Sweden University, Faculty of Science, Technology and Media, Department of Computer and Electrical Engineering (2023-).
2024 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

This study investigates how to fine-tune the large language model GPTSW3 for a specific use case scenario. By using different training methods and techniques, the model has been adapted and evaluated based on its ability to generate correct answers. The research has identified the main challenges in the training process, which includes data quality, preprocessing and finding the optimal parameter settings. The study has also examined if the model’s ability to generate accurate answers is dependent on the size of the training data. The results showed that longer training periods, combined by supervised and unsupervised training, and optimization of the parameters are critical for improving the model’s ability to generate accurate answers. Future work should focus on increasing the datasets diversity and use a larger model to further improve the models ability to generate accurate answers.

Place, publisher, year, edition, pages
2024. , p. 57
Keywords [en]
GPT-SW3, Large language model, Python, Supervised learning, Natural language processing
National Category
Computer Systems
Identifiers
URN: urn:nbn:se:miun:diva-52024Local ID: DT-V24-A2-001OAI: oai:DiVA.org:miun-52024DiVA, id: diva2:1887052
Subject / course
Computer Engineering DT1
Educational program
Master of Science in Engineering - Computer Engineering TDTEA 300 higher education credits
Supervisors
Examiners
Available from: 2024-08-06 Created: 2024-08-06 Last updated: 2024-08-06Bibliographically approved

Open Access in DiVA

fulltext(642 kB)96 downloads
File information
File name FULLTEXT01.pdfFile size 642 kBChecksum SHA-512
b12b57416862459e8efa776255c2aff8b09437160d9b5cf8871df785f017afaf843ab92de246c10f90f94a9d1aa24d237d1c4fc2acb1ba90b73c4f548ca1ea9e
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Borgström, Oliver
By organisation
Department of Computer and Electrical Engineering (2023-)
Computer Systems

Search outside of DiVA

GoogleGoogle Scholar
Total: 96 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 137 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf