Complete upload of the Urdu Multi-Domain Benchmark dataset.
- 33 datasets across Nastaliq and Roman Urdu scripts
- 284,169 total examples
- 4 task types: Sentiment Analysis, Hate Speech/Cyber Abuse, Fake News Detection, QA Pair Validation
- 1002 data files + rich Dataset Card
- Includes standardized train/test splits and original source files
