Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model

Yeskendir Koishekenov, Alexandre Berard, Vassilina Nikoulina


Abstract
The recently released NLLB-200 is a set of multilingual Neural Machine Translation models that cover 202 languages. The largest model is based on a Mixture of Experts architecture and achieves SoTA results across many language pairs. It contains 54.5B parameters and requires at least four 32GB GPUs just for inference. In this work, we propose a pruning method that enables the removal of up to 80% of experts without further finetuning and with a negligible loss in translation quality, which makes it feasible to run the model on a single 32GB GPU. Further analysis suggests that our pruning metrics can identify language-specific experts.
Anthology ID:
2023.acl-long.198
Volume:
Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Anna Rogers, Jordan Boyd-Graber, Naoaki Okazaki
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3567–3585
Language:
URL:
https://aclanthology.org/2023.acl-long.198
DOI:
10.18653/v1/2023.acl-long.198
Bibkey:
Cite (ACL):
Yeskendir Koishekenov, Alexandre Berard, and Vassilina Nikoulina. 2023. Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3567–3585, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model (Koishekenov et al., ACL 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.acl-long.198.pdf
Video:
 https://aclanthology.org/2023.acl-long.198.mp4