--- dataset_info: features: - name: corpus dtype: string splits: - name: train num_bytes: 5015779 num_examples: 19963 download_size: 3150729 dataset_size: 5015779 configs: - config_name: default data_files: - split: train path: data/train-* ---