Experimental dataset, experimental model source code, inference sentence data

MultiRC includes training and development sets, and data with different hops from the source HotPotQA dataset has been extracted.The folder 'Evidence Sentence Data Retrieved by ISR' is divided into training and testing sets for correct inference sentences and inference sentences retrieved by ISR on two datasets, facilitating model training.The folder [Deep Model Algorithm] is the source code for deep learning models, namely text classification and text extraction.