Models trained on human translated datasets to help evaluate the accuracy of machine translated texts.