GAND (Gender-Ambiguous Natural Data) dataset, a benchmarking resource for evaluating gender in machine translation.