Miðeind's WMT 2021 submission

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

We present Miðeind's submission for the English→Icelandic and Icelandic→English subsets of the 2021 WMT news translation task. Transformer-base models are trained for translation on parallel data to generate backtranslations iteratively. A pretrained mBART-25 model is then adapted for translation using parallel data as well as the last backtranslation iteration. This adapted pretrained model is then used to re-generate backtranslations, and the training of the adapted model is continued.

Original languageEnglish
Title of host publicationWMT 2021 - 6th Conference on Machine Translation, Proceedings
Number of pages4
PublisherAssociation for Computational Linguistics (ACL)
Publication date2021
Pages136-139
ISBN (Electronic)9781954085947
Publication statusPublished - 2021
Externally publishedYes
Event6th Conference on Machine Translation, WMT 2021 - Virtual, Online, Dominican Republic
Duration: 10 Nov 202111 Nov 2021

Conference

Conference6th Conference on Machine Translation, WMT 2021
LandDominican Republic
ByVirtual, Online
Periode10/11/202111/11/2021

Bibliographical note

Funding Information:
This project was supported by the Language Technology Programme for Icelandic 2019–2023. The programme, which is managed and coordinated by Almannarómur, is funded by the Icelandic Ministry of Education, Science and Culture.

Publisher Copyright:
© 2021 Association for Computational Linguistics

ID: 371184978