sabin1234/NEPALI-MCQ-SFT-MULTIDOMAIN-DATASET

Nepali Devanagari SFT Dataset — Final Clean Release A 100,000-row synthetic Nepali SFT dataset designed for Nepali-language instruction-following and supervised fine-tuning experiments. Release status: Final structural and Unicode validation passed for the previously identified contamination/corruption patterns. Dataset at a Glance Property Value Total rows 100,000 Total conversation messages 200,000 Human messages 100,000 GPT messages 100,000… See the full description on the dataset page: https://huggingface.co/datasets/sabin1234/NEPALI-MCQ-SFT-MULTIDOMAIN-DATASET.