Multilingual Non-Autoregressive Machine Translation without Knowledge Distillation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Huang, Chenyang, Huang, Fei, Zheng, Zaixiang, Zaïane, Osmar R., Zhou, Hao, Mou, Lili
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866929702143262720
author Huang, Chenyang
Huang, Fei
Zheng, Zaixiang
Zaïane, Osmar R.
Zhou, Hao
Mou, Lili
author_facet Huang, Chenyang
Huang, Fei
Zheng, Zaixiang
Zaïane, Osmar R.
Zhou, Hao
Mou, Lili
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2502_04537
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multilingual Non-Autoregressive Machine Translation without Knowledge Distillation
Huang, Chenyang
Huang, Fei
Zheng, Zaixiang
Zaïane, Osmar R.
Zhou, Hao
Mou, Lili
Computation and Language
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.
title Multilingual Non-Autoregressive Machine Translation without Knowledge Distillation
topic Computation and Language
url https://arxiv.org/abs/2502.04537