Machine Translation on Translation (test)
29.58BLEUCVT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CVTSupervision Type=Semi-supervised, Learning Paradigm=Single-task, Model Scale=Standard2018.09 | 29.58 | — | |
| ELMoSupervision Type=Semi-supervised, Learning Paradigm=Single-task, Model Scale=Standard2018.09 | 29.34 | — | |
| Word DropoutSupervision Type=Semi-supervised, Learning Paradigm=Single-task, Model Scale=Standard2018.09 | 29.33 | — | |
| SupervisedSupervision Type=Supervised, Learning Paradigm=Single-task, Model Scale=Standard2018.09 | 28.88 | — | |
| LightMoERatio=40%, # Params=0.96B2026.03 | 27.7 | 50.9 | |
| MC-SMoERatio=40%, # Params=0.96B2026.03 | 27.5 | 48.4 | |
| MoBERatio=30%, # Params=0.96B2026.03 | 27 | 53.8 | |
| HC-SMoERatio=30%, # Params=0.96B2026.03 | 26.9 | 49.8 | |
| MC-SMoERatio=30%, # Params=0.96B2026.03 | 26.8 | 49.3 | |
| HC-SMoERatio=40%, # Params=0.96B2026.03 | 26.5 | 48.9 | |
| GoogleSupervision Type=Supervised2018.09 | 26.1 | — | |
| HC-SMoERatio=50%, # Params=0.96B2026.03 | 25.8 | 45.8 | |
| LightMoERatio=30%, # Params=0.96B2026.03 | 25.3 | 51.4 | |
| MoBERatio=40%, # Params=0.96B2026.03 | 24.5 | 49 | |
| LightMoERatio=50%, # Params=0.96B2026.03 | 24 | 48.4 | |
| StanfordSupervision Type=Supervised2018.09 | 23.3 | — | |
| MC-SMoERatio=50%, # Params=0.96B2026.03 | 22.9 | 40.4 | |
| LoRARatio=0%, # Params=0.96B2026.03 | 19.9 | 52 | |
| MoBERatio=50%, # Params=0.96B2026.03 | 19.9 | 38.6 | |
| OriginalRatio=0%, # Params=-2026.03 | 12.2 | 26.2 |