prikaz prve stranice dokumenta Inženjering upita za velike jezične modele
  Download
PDF 4.61 MB
master's thesis
Inženjering upita za velike jezične modele
2024. urn:nbn:hr:200:217763
Martinović, Ivan
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Software Engineering
Chair of Visual Computing

Cite this document

Martinović, I. (2024). Inženjering upita za velike jezične modele (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek. Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:217763

Martinović, Ivan. "Inženjering upita za velike jezične modele." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2024. https://urn.nsk.hr/urn:nbn:hr:200:217763

Martinović, Ivan. "Inženjering upita za velike jezične modele." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2024. https://urn.nsk.hr/urn:nbn:hr:200:217763

Martinović, I. (2024). 'Inženjering upita za velike jezične modele', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 31 December 2024, https://urn.nsk.hr/urn:nbn:hr:200:217763

Martinović I. Inženjering upita za velike jezične modele [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2024 [cited 2024 December 31] Available at: https://urn.nsk.hr/urn:nbn:hr:200:217763

I. Martinović, "Inženjering upita za velike jezične modele", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2024. Available at: https://urn.nsk.hr/urn:nbn:hr:200:217763

Please login to the repository to save this object to your list.
Metadata
TitleInženjering upita za velike jezične modele
Title (english)Prompt Engineering for Large Language Models
AuthorIvan Martinović
MentorHrvoje Leventić (mentor)
Committee memberHrvoje Leventić (predsjednik povjerenstva)
Committee memberKrešimir Romić (član povjerenstva)
Committee memberMarin Benčević (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
(Department of Software Engineering)
(Chair of Visual Computing)
Osijek
Defense date and country2024-10-14, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
Ovaj rad bavi se primjenom i razvojem tehnika obrade prirodnog jezika kroz upotrebu velikih jezičnih modela poput GPT-a, BERT-a i drugih naprednih modela temeljenih na arhitekturi transformera. Glavni problem koji rad obrađuje je kontrola generiranja teksta i osiguranje da izlaz modela zadovoljava specifične zahtjeve i smanjuje pristranost. Kao smjernica za rješavanje ovog problema uveden je koncept inženjeringa upita, koji omogućava korisnicima da kreiraju specifične upite kako bi precizno usmjerili modele prema željenim izlazima. U radu su opisane različite metode i tehnike inženjeringa upita, uključujući dizajniranje i prilagodbu uputa modelu, te je istražena njihova učinkovitost u kontroliranju rezultata. Također su analizirani napredni modeli, poput GPT-3/4, Llama, Chinchilla i Gemini. Rezultati rada pokazuju da pažljivo oblikovani upiti mogu značajno poboljšati kvalitetu generiranih tekstova, smanjiti pristranost te omogućiti bolje korisničko iskustvo u interakcijama s AI sustavima.
Abstract (english)
This paper deals with the application and development of natural language processing techniques through the usage of large language models such as GPT, BERT, and other advanced models based on transformer architecture. The main problem addressed in the paper is the control of text generation and ensuring that the model's output meets specific requirements and reduces bias. As a guideline for solving this problem, the concept of prompt engineering is introduced, which allows users to create specific prompts to precisely direct models toward desired outputs. The paper describes various methods and techniques of prompt engineering, including the design and adaptation of instructions to the model, and examines their effectiveness in controlling results. Advanced models such as GPT-3/4, Llama, Chinchilla, and Gemini are also analyzed. The results of the study show that carefully crafted prompts can significantly improve the quality of generated texts, reduce bias, and provide a better user experience in interactions with AI systems.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:217763
Study programmeTitle: Graduate University Study Programme in Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka računarstva (sveučilišni/a magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2024-10-31 07:41:09