COSTRA 1.0: A Dataset of Complex Sentence Transformations

COSTRA 1.0 is a dataset of Czech complex sentence transformations. The dataset is intended for the study of sentence-level embeddings beyond simple word alternations or standard...