Machine Translation on CoMMuTE En-De (Accuracy)
59AccuracyVGAMT
Evaluation results
| Method | Links | |
|---|---|---|
| VGAMTObjectives=MMT + VMLM, trainable_params=13.2M2022.12 | 59 | |
| Vanilla MT*Objectives=NMT, trainable_params=4.1M2022.12 | 50 | |
| TLM + MT*Objectives=NMT, trainable_params=42M2022.12 | 50 | |
| mBART + MT*Objectives=NMT, trainable_params=-2022.12 | 50 | |
| mBART + MT* w/ adaptersObjectives=NMT + MLM, trainable_params=12.6M2022.12 | 50 | |
| VTLM + MMT*Objectives=MMT, trainable_params=44M2022.12 | 50 | |
| Gated Fusion*Objectives=MMT, trainable_params=2.8M2022.12 | 49.7 | |
| Graph-MMT*Objectives=MMT, trainable_params=4.1M2022.12 | 49.1 |