nassimjp/zamai-pashto-clean-cpt

ZamAI Pashto Clean CPT This dataset is a hyper-cleaned, optimized, and fully deduplicated version of tasal9/ZamAI-Pashto-Mega-Dataset intended for Causal Language Modeling (CLM), pre-training, or fine-tuning text models in the Pashto language. 🛠️ Pipeline & Filtering Details Before processing the ~1.5 GB stream, an Internal Built-In Self-Test (BIST) was executed to verify environment I/O permissions, validate character ratio logic, and check connection… See the full description on the dataset page: https://huggingface.co/datasets/nassimjp/zamai-pashto-clean-cpt.