Urdu Summarization Dataset containining 76,637 records of Article + Summary pairs scrapped from BBC Urdu and DW Urdu News Websites. Preprocessed Version: upto 512 tokens (~words); removed URLs, Pic Captions etc

Supported Tasks and Leaderboards

Summarization: Extractive and Abstractive

urT5 adapted from mT5 having monolingual vocabulary only; 40k tokens of Urdu. Fine-tuned version @… See the full description on the dataset page: https://huggingface.co/datasets/mbshr/XSUMUrdu-DW_BBC.