Explorations into adversarial losses on top of autoregressive loss for language modeling - View it on GitHub