Leveraging Estonian Olympiad Problems for Evaluating Large Language Models
Organization
TartuNLP
Abstract
Estonian Olympiad problems present a valuable data source for evaluating the performance of Large Language Models (LLMs). This thesis involves identifying relevant Olympiad tasks, collecting and processing data into benchmarks, and evaluating both open-source and commercial LLMs. The outcomes will provide insights into the models' capabilities in handling complex and possibly multimodal tasks in the Estonian language.
Graduation Theses defence year
2024-2025
Supervisor
Taido Purason, Hele-Andra Kuulmets
Spoken language (s)
Estonian, English
Requirements for candidates
Familiarity with machine learning and NLP. Experience with website/PDF scraping, LLM inference, and the HuggingFace ecosystem will be useful.
Level
Bachelor, Masters
Keywords
Application of contact
Name
Taido Purason
Phone
E-mail