prikaz prve stranice dokumenta Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik
No public access
master's thesis
Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik
2022. urn:nbn:hr:168:950977
Petričević, Marin
University of Zagreb
Faculty of Electrical Engineering and Computing

Cite this document

Petričević, M. (2022). Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:950977

Petričević, Marin. "Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:950977

Petričević, Marin. "Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:950977

Petričević, M. (2022). 'Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 12 April 2024, https://urn.nsk.hr/urn:nbn:hr:168:950977

Petričević M. Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2022 [cited 2024 April 12] Available at: https://urn.nsk.hr/urn:nbn:hr:168:950977

M. Petričević, "Primjena i vrednovanje velikih jezičnih modela za hrvatski jezik", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2022. Available at: https://urn.nsk.hr/urn:nbn:hr:168:950977

Please login to the repository to save this object to your list.
Metadata
TitlePrimjena i vrednovanje velikih jezičnih modela za hrvatski jezik
Title (english)Application and evaluation of large language models for the Croatian language
AuthorMarin Petričević
MentorJan Šnajder (mentor)
Committee memberJan Šnajder (predsjednik povjerenstva)
Committee memberMarko Đurasević (član povjerenstva)
Committee memberAlan Jović (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2022-07-12, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Ovaj rad uvodi veliki skup podataka sastavljen of hrvatskih novinskih članaka i predstavlja dva jezična modela za Hrvatski bazirana na arhitekutri neuronskih mreža "Transformer". Jedan je model temeljen na zadatku maskiranog jezičnog modeliranja, a drugi na zadatku kauzalnog jezičnog modeliranja. Performance ovih modela vrednovane su na nizvodnim zadacima: analizi sentimenta, prepoznavanju imenovanih entiteta i razumijevanju prirodnog jezika. Konačno, empirijski su vrednovane metode povećanja propusnosti modela, podrezivanje i kvantizacija.
Abstract (english)
This thesis introduces a new large data set of Croatian news articles and presents two Croatian language models based on the Transformer neural network architecture. One model is based on the masked language modeling task, and the other on the causal language modeling task. The performance of these models is then evaluated on downstream tasks: sentiment analysis, named entity recognition, and natural language understanding. Finally, methods of increasing model throughput, pruning and quantization, are empirically tested.
Keywords
Keywords (english)
Languageenglish
URN:NBNurn:nbn:hr:168:950977
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2023-01-13 10:35:42