Neural Machine Translation of Rare Words with Subword Units

Rico Sennrich, Barry Haddow, Alexandra Birch


Anthology ID:
P16-1162
Volume:
Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
August
Year:
2016
Address:
Berlin, Germany
Editors:
Katrin Erk, Noah A. Smith
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1715–1725
Language:
URL:
https://preview.aclanthology.org/acl-awards-reasoning/P16-1162/
DOI:
10.18653/v1/P16-1162
Award:
 ACL 2026 Test of Time Award
This work addressed the open vocabulary challenge in neural machine translation by introducing subword tokenization as a simple, practical solution. It enabled neural systems to handle rare and unseen words effectively and became the default tokenization method across NLP and beyond, from NMT to pre-training frameworks like BERT, GPT, etc. Its influence has far exceeded its original scope: Byte Pair Encoding now serves as the default tokenization method for most large language models, with impact extending beyond NLP to a wide range of domains. It exemplifies how pragmatic simplicity can become foundational.
Bibkey:
Cite (ACL):
Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. Neural Machine Translation of Rare Words with Subword Units. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1715–1725, Berlin, Germany. Association for Computational Linguistics.
Cite (Informal):
Neural Machine Translation of Rare Words with Subword Units (Sennrich et al., ACL 2016)
Copy Citation:
PDF:
https://preview.aclanthology.org/acl-awards-reasoning/P16-1162.pdf
Software:
 P16-1162.Software.zip