Semantic Similarity Cross-lingual same language XL s.
0.815Pearson's rVMSST
Evaluation results
| Method | Links | |
|---|---|---|
| VMSSTInitialization=Pretrained, Number of Layers=242022.12 | 0.815 | |
| BITRANSLATIONInitialization=Pretrained, Number of Layers=242022.12 | 0.798 | |
| VMSSTInitialization=Random, Number of Layers=242022.12 | 0.777 | |
| VMSST CONTRASTIVEInitialization=Pretrained, Number of Layers=242022.12 | 0.767 | |
| CONTRASTIVEInitialization=Pretrained, Number of Layers=242022.12 | 0.76 | |
| VMSST CONTRASTIVEInitialization=Random, Number of Layers=242022.12 | 0.755 | |
| VMSSTInitialization=Random, Number of Layers=62022.12 | 0.751 | |
| BITRANSLATIONInitialization=Random, Number of Layers=242022.12 | 0.746 | |
| CONTRASTIVEInitialization=Random, Number of Layers=62022.12 | 0.733 | |
| VMSST CONTRASTIVEInitialization=Random, Number of Layers=62022.12 | 0.73 | |
| BITRANSLATIONInitialization=Random, Number of Layers=62022.12 | 0.716 | |
| CONTRASTIVEInitialization=Random, Number of Layers=242022.12 | 0.716 |