Name
Title | ||
|---|---|---|
Learning to Evaluate Translation Beyond English: BLEURT Submissions to the WMT Metrics 2020 Shared Task |
Abstract | ||
|---|---|---|
The quality of machine translation systems has dramatically improved over the last decade, and as a result, evaluation has become an increasingly challenging problem. This paper describes our contribution to the WMT 2020 Metrics Shared Task, the main benchmark for automatic evaluation of translation. Our submission is based on BLEURT, a previously published metric based on transfer learning. We extend the metric beyond English and evaluate it on 12 languages for which training examples are available, as well as four "zero-shot" languages, for which we have no fine-tuning data. Additionally, we focus on English to German and demonstrate how to combine BLEURT's predictions with those of YiSi and use alternative reference translations to enhance the performance. Empirical results show that BLEURT achieves competitive results on the WMT Metrics 2019 Shared Task, indicating its promise for the 2020 edition. |
Year | Venue | DocType |
|---|---|---|
2020 | WMT@EMNLP | Conference |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
8 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Thibault Sellam | 1 | 37 | 9.53 |
| Amy Pu | 2 | 0 | 0.34 |
| Hyung Won Chung | 3 | 0 | 3.04 |
| Sebastian Gehrmann | 4 | 84 | 10.58 |
| Qijun Tan | 5 | 0 | 0.68 |
| Markus Freitag | 6 | 86 | 15.28 |
| Dipanjan Das | 7 | 1619 | 75.14 |
| Ankur P. Parikh | 8 | 250 | 18.47 |