Machine Translation on AmericasNLP 2023 (dev)
36.24chrF++ (AYM)Sheffield
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SheffieldModel=NLLB-200, Augmentation=back-translation2026.01 | 36.24 | 39.34 | 39.52 | |
| Helsinki-NLP2026.01 | 33.44 | 40.42 | 37.19 | |
| mBART (Ours)Data=Curated + Synthetic2026.01 | 30.82 | 44 | 37.83 | |
| PlayGround2026.01 | 29.98 | 33.17 | 34.38 | |
| 2021 Best System2026.01 | 28.3 | 33.6 | 34.3 | |
| mBART (Ours)Data=Curated-only2026.01 | 26.85 | 42 | 37.12 | |
| LTLAmsterdam2026.01 | 25.23 | 32.89 | 36.81 | |
| CIC-NLP2026.01 | 19.05 | 21.75 | 35.62 | |
| 2021 Baselinereference=Vázquez et al., 20212026.01 | 15.7 | 19.3 | 30.4 | |
| Andes2026.01 | 9.22 | — | — | |
| LCT-EHU2026.01 | — | — | 38.59 |