Abstract
This paper explores knowledge distillation for multi-domain neural machine translation (NMT). We focus on the Estonian-English translation direction and experiment with distilling the knowledge of multiple domain-specific teacher models into a single student model that is tiny and efficient. Our experiments use a large parallel dataset of 18 million sentence pairs, consisting of 10 corpora, divided into 6 domain groups based on source similarity, and incorporate forward-translated monolingual data. Results show that tiny student models can cope with multiple domains even in case of large corpora, with different approaches benefiting frequent and low-resource domains.- Anthology ID:
- 2023.nodalida-1.78
- Volume:
- Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)
- Month:
- May
- Year:
- 2023
- Address:
- Tórshavn, Faroe Islands
- Editors:
- Tanel Alumäe, Mark Fishel
- Venue:
- NoDaLiDa
- SIG:
- Publisher:
- University of Tartu Library
- Note:
- Pages:
- 772–781
- Language:
- URL:
- https://aclanthology.org/2023.nodalida-1.78
- DOI:
- Cite (ACL):
- Elizaveta Korotkova and Mark Fishel. 2023. Distilling Estonian Text Domains for Production-Oriented Machine Translation. In Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa), pages 772–781, Tórshavn, Faroe Islands. University of Tartu Library.
- Cite (Informal):
- Distilling Estonian Text Domains for Production-Oriented Machine Translation (Korotkova & Fishel, NoDaLiDa 2023)
- PDF:
- https://preview.aclanthology.org/emnlp22-frontmatter/2023.nodalida-1.78.pdf