Institutional Repository of Vilnius University
Chat GPT ir Mistral AI panaudojimo palyginimas klausimų generavimui iš pateikto teksto /
Abstract
dc:descriptionThe main aim of the thesis is to compare different language models to generate questions from a given text. This thesis compares the ChatGPT 3.5 and Mistral 7B language models.The thesis consists of five parts: an overview of language models; opportunities and challenges of using language models in education; requirements for the system, prototype and system development; comparison of the results of the question generation with language models; system evaluation with educational content development models. During the literature review, the advantages of language models in the field of education became apparent – language models can have a positive influence on the development of students’ reading, writing and text comprehension, help disabled students to get involved in learning processes, help teachers to do their work more efficiently, and generate ideas for tests. Challenges such as copyright infringement, bias and possible promotion of fraud are also highlighted. Evaluating the test results of the developed system, it can be observed that the system using ChatGPT 3.5 worked faster, used less system resources and generated higher quality questions. Meanwhile, Mistral 7B has other advantages – during testing phase it worked more stably, is free of charge and more secure in terms of data protection. Based on the results of the conducted survey and the TAM evaluation model of the systems, it can be concluded that educators would benefit from such system and it’s use would be easy to understand.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Brazauskas, Kernius,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210645296&prefLang=en_US