Published in the Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL 2025), Mumbai, India.
The paper presents FarSense, a Farsi commonsense benchmark with six tasks (true/false, multiple choice, explanation, cause-effect, counterfactual reasoning and knowledge completion) built from about 4,210 Farsi Wikipedia passages and validated by native speakers. It evaluates five major LLMs and shows that fine-tuning six compact models (1B to 24B) on FarSense brings substantial gains.
Authors: Kamyar Zeinalipour, Neda Jamshidi, Seyedehbahareh Hejazi, Marco Maggini, Monica Bianchini, Simone Paoletti, Marco Gori.
Read the paper | Code | Models and data
All the lab's language releases are listed in Language models and datasets.