Non-autoregressive machine translation (NAT) has recently made great progress. However, most works to date have focused on standard translation tasks, even though some edit-based NAT models, such as the Levenshtein Transformer (LevT), seem well suited to translate with a Translation Memory (TM). This is the scenario considered here. We first analyze the vanilla LevT model and explain why it does not do well in this setting. We then propose a new variant, TM-LevT, and show how to effectively train this model. By modifying the data presentation and introducing an extra deletion operation, we obtain performance that are on par with an autoregressive approach, while reducing the decoding load. We also show that incorporating TMs during training dispenses to use knowledge distillation, a well-known trick used to mitigate the multimodality issue.

翻译：非自回归机器翻译（NAT）近年来取得了显著进展。然而，现有研究大多集中于标准翻译任务，尽管诸如Levenshtein Transformer（LevT）等基于编辑的NAT模型似乎非常适合在翻译记忆库（TM）辅助下进行翻译。本文正是针对这一场景展开研究。我们首先分析了原始LevT模型，并解释了其在此场景中表现不佳的原因。随后提出一种新变体——TM-LevT，并展示了如何有效训练该模型。通过调整数据呈现方式并引入额外的删除操作，我们获得了与自回归方法相当的性能，同时降低了解码负载。此外，我们还发现，在训练过程中引入翻译记忆库可避免使用知识蒸馏——这一常用于缓解多模态问题的经典技巧。