Abstract
Transfer learning improves quality for low-resource machine translation, but it is unclear what exactly it transfers. We perform several ablation studies that limit information transfer, then measure the quality impact across three language pairs to gain a black-box understanding of transfer learning. Word embeddings play an important role in transfer learning, particularly if they are properly aligned. Although transfer learning can be performed without embeddings, results are sub-optimal. In contrast, transferring only the embeddings but nothing else yields catastrophic results. We then investigate diagonal alignments with auto-encoders over real languages and randomly generated sequences, finding even randomly generated sequences as parents yield noticeable but smaller gains. Finally, transfer learning can eliminate the need for a warm-up phase when training transformer models in high resource language pairs.
Original language | English |
---|---|
Title of host publication | Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics |
Publisher | Association for Computational Linguistics (ACL) |
Pages | 7701–7710 |
Number of pages | 10 |
ISBN (Electronic) | 978-1-952148-25-5 |
DOIs | |
Publication status | Published - 10 Jul 2020 |
Event | 2020 Annual Conference of the Association for Computational Linguistics - Hyatt Regency Seattle, Virtual conference, United States Duration: 5 Jul 2020 → 10 Jul 2020 Conference number: 58 https://acl2020.org/ |
Conference
Conference | 2020 Annual Conference of the Association for Computational Linguistics |
---|---|
Abbreviated title | ACL 2020 |
Country/Territory | United States |
City | Virtual conference |
Period | 5/07/20 → 10/07/20 |
Internet address |