Building Multilingual Machine Translation Systems That Serve Arbitrary X-Y Translations
This addresses the challenge of building cost-effective multilingual translation systems that serve arbitrary translation directions for deployment in practical scenarios.
The paper tackled the problem of poor performance in one-to-many and many-to-many translation directions in Multilingual Neural Machine Translation by proposing a two-stage training strategy of pretraining and finetuning, resulting in systems that outperform conventional baselines by an average of +6.0 and +4.1 BLEU on the WMT'21 task.
Multilingual Neural Machine Translation (MNMT) enables one system to translate sentences from multiple source languages to multiple target languages, greatly reducing deployment costs compared with conventional bilingual systems. The MNMT training benefit, however, is often limited to many-to-one directions. The model suffers from poor performance in one-to-many and many-to-many with zero-shot setup. To address this issue, this paper discusses how to practically build MNMT systems that serve arbitrary X-Y translation directions while leveraging multilinguality with a two-stage training strategy of pretraining and finetuning. Experimenting with the WMT'21 multilingual translation task, we demonstrate that our systems outperform the conventional baselines of direct bilingual models and pivot translation models for most directions, averagely giving +6.0 and +4.1 BLEU, without the need for architecture change or extra data collection. Moreover, we also examine our proposed approach in an extremely large-scale data setting to accommodate practical deployment scenarios.