Towards Mitigating Gender Bias in a decoder-based Neural Machine Translation model by Adding Contextual Information