Scaling Low-Resource MT via Synthetic Data Generation with LLMs
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914050090205184 |
|---|---|
| author | de Gibert, Ona Attieh, Joseph Vahtola, Teemu Aulamo, Mikko Li, Zihao Vázquez, Raúl Hu, Tiancheng Tiedemann, Jörg |
| author_facet | de Gibert, Ona Attieh, Joseph Vahtola, Teemu Aulamo, Mikko Li, Zihao Vázquez, Raúl Hu, Tiancheng Tiedemann, Jörg |
| contents | We investigate the potential of LLM-generated synthetic data for improving low-resource Machine Translation (MT). Focusing on seven diverse target languages, we construct a document-level synthetic corpus from English Europarl, and extend it via pivoting to 147 additional language pairs. Automatic and human evaluation confirm its overall high quality. We study its practical application by (i) identifying effective training regimes, (ii) comparing our data with the HPLT dataset, (iii) studying the effect of varying training data size, and (iiii) testing its utility beyond English-centric MT. Finally, we introduce SynOPUS, a public repository for synthetic parallel datasets. Our findings show that LLM-generated synthetic data, even when noisy, can substantially improve MT performance for low-resource languages. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_14423 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Scaling Low-Resource MT via Synthetic Data Generation with LLMs de Gibert, Ona Attieh, Joseph Vahtola, Teemu Aulamo, Mikko Li, Zihao Vázquez, Raúl Hu, Tiancheng Tiedemann, Jörg Computation and Language We investigate the potential of LLM-generated synthetic data for improving low-resource Machine Translation (MT). Focusing on seven diverse target languages, we construct a document-level synthetic corpus from English Europarl, and extend it via pivoting to 147 additional language pairs. Automatic and human evaluation confirm its overall high quality. We study its practical application by (i) identifying effective training regimes, (ii) comparing our data with the HPLT dataset, (iii) studying the effect of varying training data size, and (iiii) testing its utility beyond English-centric MT. Finally, we introduce SynOPUS, a public repository for synthetic parallel datasets. Our findings show that LLM-generated synthetic data, even when noisy, can substantially improve MT performance for low-resource languages. |
| title | Scaling Low-Resource MT via Synthetic Data Generation with LLMs |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2505.14423 |