This data is a subset of the BioQA task B dataset. It includes only factoid samples for extractive QA and is split into train and test with 80% and 20% respectively.