Natural Language Inference on HANS Unknown Bias (out-of-distribution)
70.7AccuracyPoEDirichlet
Evaluation Results
| Method | Links | |
|---|---|---|
| PoEDirichletCalibrator=Dirichlet2021.11 | 70.7 | |
| DRiFtTempsCalibrator=Temperature Scaling2021.11 | 70 | |
| PoETempsCalibrator=Temperature Scaling2021.11 | 69.9 | |
| DRiFtDirichletCalibrator=Dirichlet2021.11 | 69.4 | |
| DRiFtCalibrator=None2021.11 | 69.1 | |
| PoECalibrator=None2021.11 | 69 | |
| InvRDirichletCalibrator=Dirichlet2021.11 | 68.2 | |
| LMinCalibrator=None2021.11 | 66.5 | |
| LMinDirichletCalibrator=Dirichlet2021.11 | 66.4 | |
| LMinTempsCalibrator=Temperature Scaling2021.11 | 66.2 | |
| InvRTempsCalibrator=Temperature Scaling2021.11 | 65.8 | |
| InvRCalibrator=None2021.11 | 64.7 | |
| CECalibrator=None2021.11 | 61.2 |