Multilingual instruction-tuning dataset with Arabic and French coverage.
Open benchmark dataset for Moroccan Arabic (Darija) NLP tasks.
158k Darija/MSA article-headline pairs from Goud.ma for text summarization.