Document Question Answering Datasets
There are 6 document question answering datasets in our directory. Each links to its source, paper, and download — browse the full list below or filter by language.
Document Question Answering is the task of answering questions grounded in a document, often using its layout and structure. We catalog 6 datasets for it.
- AmazonScience/document-haystackQuestion Answering, Visual Question Answering, Document Question AnsweringEN
- Eioss/CC-OCR-V2Image To Text, Visual Question Answering, Document Question AnsweringEN, ZH
- mawadalla/scientific-figures-captions-contextVisual Question Answering, Document Question AnsweringEN
- princeton-nlp/CharXivVisual Question Answering, Document Question Answering, Question AnsweringEN
- vidore/colpali_train_setDocument Question Answering, Visual Document RetrievalEnglish
- VLR-CVC/DocVQA-2026Visual Question Answering, Document Question Answering, Image Text To Text, Question AnsweringEN