LLMzSzŁ: a comprehensive LLM benchmark for Polish
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929658104119296 |
|---|---|
| author | Jassem, Krzysztof Ciesiółka, Michał Graliński, Filip Jabłoński, Piotr Pokrywka, Jakub Kubis, Marek Jabłońska, Monika Staruch, Ryszard |
| author_facet | Jassem, Krzysztof Ciesiółka, Michał Graliński, Filip Jabłoński, Piotr Pokrywka, Jakub Kubis, Marek Jabłońska, Monika Staruch, Ryszard |
| contents | This article introduces the first comprehensive benchmark for the Polish language at this scale: LLMzSzŁ (LLMs Behind the School Desk). It is based on a coherent collection of Polish national exams, including both academic and professional tests extracted from the archives of the Polish Central Examination Board. It covers 4 types of exams, coming from 154 domains. Altogether, it consists of almost 19k closed-ended questions. We investigate the performance of open-source multilingual, English, and Polish LLMs to verify LLMs' abilities to transfer knowledge between languages. Also, the correlation between LLMs and humans at model accuracy and exam pass rate levels is examined. We show that multilingual LLMs can obtain superior results over monolingual ones; however, monolingual models may be beneficial when model size matters. Our analysis highlights the potential of LLMs in assisting with exam validation, particularly in identifying anomalies or errors in examination tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_02266 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LLMzSzŁ: a comprehensive LLM benchmark for Polish Jassem, Krzysztof Ciesiółka, Michał Graliński, Filip Jabłoński, Piotr Pokrywka, Jakub Kubis, Marek Jabłońska, Monika Staruch, Ryszard Computation and Language Artificial Intelligence This article introduces the first comprehensive benchmark for the Polish language at this scale: LLMzSzŁ (LLMs Behind the School Desk). It is based on a coherent collection of Polish national exams, including both academic and professional tests extracted from the archives of the Polish Central Examination Board. It covers 4 types of exams, coming from 154 domains. Altogether, it consists of almost 19k closed-ended questions. We investigate the performance of open-source multilingual, English, and Polish LLMs to verify LLMs' abilities to transfer knowledge between languages. Also, the correlation between LLMs and humans at model accuracy and exam pass rate levels is examined. We show that multilingual LLMs can obtain superior results over monolingual ones; however, monolingual models may be beneficial when model size matters. Our analysis highlights the potential of LLMs in assisting with exam validation, particularly in identifying anomalies or errors in examination tasks. |
| title | LLMzSzŁ: a comprehensive LLM benchmark for Polish |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2501.02266 |