Machine Translation on WMT'14 English-German (test)
27.27BLEULevenshtein Transformer
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Levenshtein TransformerTraining teacher=distillation (autoregressive teacher model), Model architecture=Transformer base2019.05 | 27.27 | 92 | 2.05 | |
| TransformerDecoding strategy=beam-search (beam=4), Model architecture=Transformer base2019.05 | 27.17 | 369 | 28.1 | |
| TransformerDecoding strategy=greedy, Model architecture=Transformer base2019.05 | 26.89 | 343 | 28.1 | |
| Levenshtein TransformerTraining teacher=oracle, Model architecture=Transformer base2019.05 | 25.2 | 126 | 2.88 |