This paper introduces Saudi-Dialect-ALLaM, a LoRA fine-tuned version of the Saudi Arabian foundation model ALLaM-7B-Instruct-preview, designed to improve the generation of Saudi dialects (Najdi and Hijazi). The model is trained on a private dataset of 5,466 synthetic instruction-response pairs, with two variants explored: Dialect-Token and No-Token training. Results indicate that the Dialect-Token model achieves superior dialect control and fidelity compared to generic instruction models, although the dataset and model weights are not released.
This paper introduces Absher, a new benchmark for evaluating LLMs' linguistic and cultural competence in Saudi dialects. The benchmark comprises over 18,000 multiple-choice questions spanning six categories, using dialectal words, phrases, and proverbs from various regions of Saudi Arabia. Evaluation of state-of-the-art LLMs reveals performance gaps, especially in cultural inference and contextual understanding, highlighting the need for dialect-aware training.
The paper introduces Aladdin-FTI, a system designed for generating and translating dialectal Arabic (DA). Aladdin-FTI supports text generation in Moroccan, Egyptian, Palestinian, Syrian, and Saudi dialects. It also handles bidirectional translation between these dialects, Modern Standard Arabic (MSA), and English. Why it matters: This work contributes to addressing the under-representation of Arabic dialects in NLP research and enables more inclusive Arabic language models.