XCrowd: Combining Explainability and Crowdsourcing to Diagnose Models in Relation Extraction Paper Data format The annotations are stored in TSV format. We include both annotations from each worker and those aggregated by the majority vote. How to read the annotations in Python