Do LLMs Understand Romanian Driving Laws? A Study on Multimodal and Fine-Tuned Question Answering
Proceedings of the ... International Conference "Linguistic Resources and Tools for Natural Language Processing" · 31.12.2025 · 222
DOI: 10.47743/consilr2025.15 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the ... International Conference "Linguistic Resources and Tools for Natural Language Processing", online erschienen am 31.12.2025. Verfasst von Eduard Barbu und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Ensuring that both new and experienced drivers master current traffic rules is critical to road safety. This paper evaluates Large Language Models (LLMs) on Romanian driving-law QA with explanation generation. We release a 1,208question dataset (387 multimodal) and compare text-only and multimodal SOTA systems, then measure the impact of domain-specific fine-tuning for Llama 3.18B-Instruct and RoLlama 3.1-8B-Instruct. SOTA models perform well, but finetuned 8B models are competitive. Textual descriptions of images outperform direct visual input. Finally, an LLM-as-a-Judge assesses explanation quality, revealing self-preference bias. The study informs explainable QA for lessresourced languages.
Zitieren
APA
Barbu, E., & Dumitran, A. M. (2025). Do LLMs Understand Romanian Driving Laws? A Study on Multimodal and Fine-Tuned Question Answering. Proceedings of the ... International Conference "Linguistic Resources and Tools for Natural Language Processing", 222. https://doi.org/10.47743/consilr2025.15
Vancouver
Barbu E, Dumitran AM. Do LLMs Understand Romanian Driving Laws? A Study on Multimodal and Fine-Tuned Question Answering. Proceedings of the ... International Conference "Linguistic Resources and Tools for Natural Language Processing". 2025:222. doi:10.47743/consilr2025.15
BibTeX
@article{Barbu2025,
author = {Barbu, Eduard and Dumitran, Adrian Marius},
title = {Do LLMs Understand Romanian Driving Laws? A Study on Multimodal and Fine-Tuned Question Answering},
journal = {Proceedings of the ... International Conference "Linguistic Resources and Tools for Natural Language Processing"},
year = {2025},
pages = {222},
doi = {10.47743/consilr2025.15}
}
EUVIMED