Self-Improvement of Non-autoregressive Model via Sequence-Level Distillation

Publication
Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing