Hint-Based Training for Non-Autoregressive Machine Translation
This addresses the problem of slow inference in machine translation for applications requiring real-time processing, representing a significant but incremental improvement over prior non-autoregressive methods.
The paper tackles the trade-off between translation accuracy and inference speed in machine translation by proposing a hint-based training approach for non-autoregressive models, achieving results comparable to a strong autoregressive baseline while being 10 times faster on WMT14 En-De and De-En datasets.
Due to the unparallelizable nature of the autoregressive factorization, AutoRegressive Translation (ART) models have to generate tokens sequentially during decoding and thus suffer from high inference latency. Non-AutoRegressive Translation (NART) models were proposed to reduce the inference time, but could only achieve inferior translation accuracy. In this paper, we proposed a novel approach to leveraging the hints from hidden states and word alignments to help the training of NART models. The results achieve significant improvement over previous NART models for the WMT14 En-De and De-En datasets and are even comparable to a strong LSTM-based ART baseline but one order of magnitude faster in inference.