Search Results for author:
Found 2 papers, 0 papers with code
Prediction Difference Regularization against Perturbation for Neural Machine Translation
Regularization methods applying input perturbation have drawn considerable attention and have been frequently explored for NMT tasks in recent years.
Guiding Teacher Forcing with Seer Forcing for Neural Machine Translation
Meanwhile, we force the conventional decoder to simulate the behaviors of the seer decoder via knowledge distillation.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
