[2010.09402] Revisiting Modularized Multilingual NMT to Meet Industrial Demandsopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

The complete sharing of parameters for multilingual translation (1-1) has been the mainstream approach in current research. However, degraded performance due to the capacity bottleneck and low maintainability hinders its extensive adoption in industries. In this study, we revisit the multilingual neural machine translation model that only share modules among the same languages (M2) as a practical alternative to 1-1 to satisfy industrial requirements. Through comprehensive experiments, we identify the benefits of multi-way training and demonstrate that the M2 can enjoy these benefits without suffering from the capacity bottleneck. Furthermore, the interlingual space of the M2 allows convenient modification of the model. By leveraging trained modules, we find that incrementally added modules exhibit better performance than singly trained models. The zero-shot performance of the added modules is even comparable to supervised models. Our findings suggest that the M2 can be a competent cand

1 mentions: @kchonyc
Date: 2020/11/18 03:51

Referring Tweets

@kchonyc t.co/W3CLS8vvXz by Lu, Son, Yang & Bae t.co/GOpFICBBUN @orf_bnw we weren't wrong but just a bit ahead of time ;) t.co/gOEo0erQKq

Related Entries

Read more Algorithms for Hyper-Parameter Optimization
0 users, 1 mentions 2020/04/16 18:52
Read more Emergent Translation in Multi-Agent Communication | OpenReview
0 users, 1 mentions 2020/06/06 21:51
Read more ICML LAOW2020 - Call For Papers
0 users, 1 mentions 2020/06/19 03:52
Read more [2007.02561] Learning from Failure: Training Debiased Classifier from Biased Classifieropen searchop...
0 users, 1 mentions 2020/07/22 15:51
Read more CVPR 2019 Open Access Repository
0 users, 1 mentions 2020/08/10 11:21
Read more [2009.07177] Iterative Refinement in the Continuous Space for Non-Autoregressive Neural Machine Tran...
0 users, 1 mentions 2020/09/16 03:51