Multilingual Non-Autoregressive Machine Translation without Knowledge Distillation
This work addresses efficiency challenges in multilingual translation for NLP practitioners, though it appears incremental as it builds on existing directed acyclic Transformer methods.
The paper tackles the problem of improving efficiency in multilingual neural machine translation by proposing a non-autoregressive approach that eliminates the need for expensive knowledge distillation, achieving state-of-the-art performance.
Multilingual neural machine translation (MNMT) aims at using one single model for multiple translation directions. Recent work applies non-autoregressive Transformers to improve the efficiency of MNMT, but requires expensive knowledge distillation (KD) processes. To this end, we propose an M-DAT approach to non-autoregressive multilingual machine translation. Our system leverages the recent advance of the directed acyclic Transformer (DAT), which does not require KD. We further propose a pivot back-translation (PivotBT) approach to improve the generalization to unseen translation directions. Experiments show that our M-DAT achieves state-of-the-art performance in non-autoregressive MNMT.