Foundational Tamazight text corpora for NLP. Curated datasets optimized for pre-training and fine-tuning language models.