data/tokenized/absa_cls_dataset/dataset_dict.json

{"splits": ["train", "validation", "test"]}