Fine-Tuning for Neural Machine Translation with Limited Degradation across In- and Out-of-Domain Data

Praveen Dakwale, Christof Monz