cheemasohail/Urdu-CLEVR

--- language: - ur task_categories: - visual-question-answering task_ids: - visual-question-answering pretty_name: Urdu-CLEVR size_categories: - 100K<n<1M license: mit --- # Urdu-CLEVR Dataset Urdu-CLEVR is a synthetic Visual Question Answering (VQA) dataset for Urdu, based on the CLEVR benchmark. ## Features - 850K question-answer pairs - Urdu questions and answers - 9 reasoning types (count, compare, exist, etc.) ## Fields - image: image path - question_ur: Urdu question - question_type: reasoning category ## Note Images are based on the CLEVR dataset and are not included here to reduce storage size. ## Usage ```python from datasets import load_dataset dataset = load_dataset("cheemasohail/urdu-clevr")