Atlas-Chat: Adapting Large Language Models for Low-Resource Moroccan Arabic Dialect

TL;DR

Atlas-Chat introduces specialized large language models for Moroccan Arabic dialect, demonstrating superior performance in instruction-following and NLP tasks through novel datasets and fine-tuning strategies, filling a gap in low-resource language modeling.

Contribution

The paper presents the first LLMs tailored for Moroccan Arabic, with new datasets, training methods, and evaluation benchmarks for low-resource dialects.

Findings

Models outperform state-of-the-art Arabic LLMs

9B model improves 13% over larger models on DarijaMMLU

Fine-tuning strategies significantly impact performance

Abstract

We introduce Atlas-Chat, the first-ever collection of LLMs specifically developed for dialectal Arabic. Focusing on Moroccan Arabic, also known as Darija, we construct our instruction dataset by consolidating existing Darija language resources, creating novel datasets both manually and synthetically, and translating English instructions with stringent quality control. Atlas-Chat-2B, 9B, and 27B models, fine-tuned on the dataset, exhibit superior ability in following Darija instructions and performing standard NLP tasks. Notably, our models outperform both state-of-the-art and Arabic-specialized LLMs like LLaMa, Jais, and AceGPT, e.g., our 9B model gains a 13% performance boost over a larger 13B model on DarijaMMLU, in our newly introduced evaluation suite for Darija covering both discriminative and generative tasks. Furthermore, we perform an experimental analysis of various fine-tuning strategies and base model choices to determine optimal configurations. All our resources are publicly accessible, and we believe our work offers comprehensive design methodologies of instruction-tuning for low-resource languages, which are often neglected in favor of data-rich languages by contemporary LLMs.