Machine Translation on TED data (test)
63.3ChrF++Supervised fine-tune
Evaluation Results
| Method | Links | |
|---|---|---|
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 63.3 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 63.3 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 62.4 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 62.4 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 61.2 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 60.4 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 60.4 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 58.5 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 58.1 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 57.9 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 57.5 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 57.3 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 56.5 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 56.4 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 56.3 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 56.1 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 56 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 55.2 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 55.2 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 54.7 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 54.4 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 54.3 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 54.3 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 53.5 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 53.2 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 52.7 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 52.6 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 52.6 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 51.9 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 51.8 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 51.2 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 50.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 50.5 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 49.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 49.4 | |
| Supervised fine-tuneArchitecture=Softmax, L_base=5, d_model=3002024.08 | 48.5 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 48.4 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 48.2 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 48.2 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 48.1 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 47.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 47.9 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 47.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 47.6 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 47.5 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 47.4 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 45.7 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 45.5 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 45.5 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 45.4 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 45.2 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 44.6 | |
| Supervised fine-tuneArchitecture=vMF, L_base=5, d_model=3002024.08 | 44.3 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 42.4 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 42.4 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 42.1 | |
| Softmax (Garcia et al., 2021)Architecture=Softmax, L_base=5, d_model=3002024.08 | 41.7 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 41.7 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 40.9 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 40.5 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 40 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 39.6 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 38.8 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 37.1 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 37.1 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 37 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 36 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 35.1 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 35.1 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 34.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 34.9 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=5762024.08 | 34.8 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 34.3 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 33.9 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 33.8 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 32.8 | |
| Multilingual SoftmaxArchitecture=Softmax, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 32.5 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 31.2 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, L_base=8, d_model=3002024.08 | 31.2 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, ru, uk, pl, vi, zh, L_base=10, d_model=3002024.08 | 31 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 30.8 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 30.1 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 30 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, es, fr, it, L_base=5, d_model=3002024.08 | 29.4 | |
| vMF-NMTArchitecture=vMF, Trained on=en, de, fr, it, L_base=4, d_model=3002024.08 | 29.2 | |
| vMF-NMTArchitecture=vMF, Trained on=es+en, L_base=2, d_model=3002024.08 | 29 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 27.5 | |
| vMF-NMTArchitecture=vMF, Trained on=es, en, L_base=1, d_model=3002024.08 | 26.9 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 23.4 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 16.1 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 12.4 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 11.4 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 11 | |
| BPE baselineArchitecture=Softmax, L_base=5, d_model=3002024.08 | 9.9 |