Evaluating Bias Detection in Lightweight LLMs
Vilnius University Open Series · 08.05.2026 · 36-39
DOI: 10.15388/lmitt.2026.4 ↗Kurz gefasst
Fachartikel in der Zeitschrift Vilnius University Open Series, online erschienen am 08.05.2026. Verfasst von Veronika Bryskina und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study evaluates the ability of lightweight open-source large language models (LLMs) to detect bias in text. Eleven models of six popular LLM families were tested in a zero-shot setting on a unified dataset of 8,745 sentences derived from three selected sources, covering gender, race, religion, and appearance bias. Results showed that none of the models exceeded 70% accuracy, which highlighted limitations of lightweight LLMs and existing challenges related to current bias detection datasets.
Zitieren
APA
Bryskina, V., Songailaitė, M., & Mandravickaitė, J. (2026). Evaluating Bias Detection in Lightweight LLMs. Vilnius University Open Series, 36-39. https://doi.org/10.15388/lmitt.2026.4
Vancouver
Bryskina V, Songailaitė M, Mandravickaitė J. Evaluating Bias Detection in Lightweight LLMs. Vilnius University Open Series. 2026:36-39. doi:10.15388/lmitt.2026.4
BibTeX
@article{Bryskina2026,
author = {Bryskina, Veronika and Songailaitė, Milita and Mandravickaitė, Justina},
title = {Evaluating Bias Detection in Lightweight LLMs},
journal = {Vilnius University Open Series},
year = {2026},
pages = {36--39},
doi = {10.15388/lmitt.2026.4}
}
EUVIMED