Papers by Anaïs Mottaz
TransBERT: A Framework for Synthetic Translation in Domain-Specific Language Modeling (2025.findings-emnlp)
Copied to clipboard
| Challenge: | TransBERT framework for pre-training language models using exclusively synthetically translated text is limited in specialized domains. |
| Approach: | They propose a framework for pre-training language models using exclusively synthetically translated text . they also introduce a scalable translation toolkit that leverages synthetically trained data . |
| Outcome: | The proposed framework can be used to train language models using synthetically translated text . transCorpus toolkit can be scalable to the life sciences domain in french . |