report
GenAI for Moroccan Darija: Challenges and Early Results
Abstract
Conference Presentation
Presented at the 7th International Congress for Moroccan Arabic, University of Navarra, Spain, 2024.
This talk discusses the unique challenges of applying generative AI to Moroccan Darija and presents early results from Omneity Labs' work on LLM training for this underrepresented language.
Citation
Related Research
Sawtone: A Universal Framework for Phonetic Similarity and Alignment Across Languages and Scripts
Introduces a cross-script phonetic alignment framework with modular language-specific adapters. Demonstrates 88% BLEU transliteration and 87–95% phonetic alignment accuracy across language/script pairs. Includes a case study on preprocessing Moroccan Arabic (Darija) for LLM training.
Gherbal: A Multilingual Classifier for Low-Resource Languages
Conference presentation at TIM'24, introducing Gherbal — a multilingual classifier designed for low-resource languages.
Gherbal v4: Comprehensive Evaluation Report — Language Identification Across 214 Languages
Comprehensive evaluation of Gherbal v4 across 10 models, 8 benchmarks, and 5 scoping regimes. Gherbal v4 achieves 0.836 average accuracy — outperforming OpenLID v2, GlotLID, and NLLB-LID. The report covers Arabic dialect identification, Arabizi detection, African language coverage, and model efficiency analysis.