unsegmented_documents: All unsegmented documents, each named with a unique id<id>.txt.segmented_documents: All segmented documents, each named with a unique id and part number<id>part<part>.txt. Documents share the same id in bothunsegmented_documentsandsegmented_documents. Thus,segmented_documents/0000part00.txtandsegmented_documents/0000part01are the first and second part respectively ofunsegmented_documents/0000.txt.qa_examples.csv: Contains the training question-answer pairs.eval_questions.csv: Validation and test question-answer pairs.
names.txt contains a list of names which is sampled during dataset generation.