Machine Translation on En-Fr 2016 (test)
67.2BLEUmBART + MT* w/ adapters
Evaluation results
| Method | Links | ||
|---|---|---|---|
| mBART + MT* w/ adaptersObjectives=NMT + MLM, trainable_params=12.6M2022.12 | 67.2 | 0.971 | |
| VGAMTObjectives=MMT + VMLM, trainable_params=13.2M2022.12 | 67.2 | 0.968 | |
| TLM + MT*Objectives=NMT, trainable_params=42M2022.12 | 62 | 0.795 | |
| VTLM + MMT*Objectives=MMT, trainable_params=44M2022.12 | 61.4 | 0.783 | |
| Vanilla MT*Objectives=NMT, trainable_params=4.0M2022.12 | 59.4 | 0.711 | |
| Graph-MMT*Objectives=MMT, trainable_params=4.0M2022.12 | 58.9 | 0.705 | |
| Gated Fusion*Objectives=MMT, trainable_params=2.8M2022.12 | 58.7 | 0.707 | |
| mBART + MT*Objectives=NMT, trainable_params=-2022.12 | 49 | 0.819 |