Metric: BLEU (higher is better)
| # | Model↕ | BLEU▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | EnViT5 + MTet | 40.2 | Yes | MTet: Multi-domain Translation for English and V... | 2022-10-11 | Code |
| 2 | Tall Transformer with Style-Augmented Training | 37.8 | Yes | - | - | Code |
| 3 | Transformer+BPE-dropout | 33.27 | No | BPE-Dropout: Simple and Effective Subword Regula... | 2019-10-29 | Code |
| 4 | Transformer+BPE+FixNorm+ScaleNorm | 32.8 | No | Transformers without Tears: Improving the Normal... | 2019-10-14 | Code |
| 5 | Transformer+LayerNorm-simple | 31.4 | No | Understanding and Improving Layer Normalization | 2019-11-16 | Code |
| 6 | CVT | 29.6 | Yes | Semi-Supervised Sequence Modeling with Cross-Vie... | 2018-09-22 | Code |
| 7 | Self-Adaptive Control of Temperature | 29.12 | No | Learning When to Concentrate or Divert Attention... | 2018-08-22 | Code |
| 8 | SAWR | 29.09 | No | Syntax-Enhanced Neural Machine Translation with ... | 2019-05-08 | - |
| 9 | DeconvDec | 28.47 | No | Deconvolution-Based Global Decoding for Neural M... | 2018-06-10 | Code |
| 10 | LSTM+Attention+Ensemble | 26.4 | No | - | - | Code |