Natural Language Inference on MNLI HardCD (challenge)
0.803AccuracyLMinTemps
Evaluation Results
| Method | Links | |
|---|---|---|
| LMinTempsCalibrator=Temperature Scaling2021.11 | 0.803 | |
| LMinCalibrator=None2021.11 | 0.8 | |
| LMinDirichletCalibrator=Dirichlet2021.11 | 0.798 | |
| PoETempsCalibrator=Temperature Scaling2021.11 | 0.796 | |
| DRiFtTempsCalibrator=Temperature Scaling2021.11 | 0.796 | |
| DRiFtDirichletCalibrator=Dirichlet2021.11 | 0.796 | |
| DRiFtCalibrator=None2021.11 | 0.795 | |
| PoECalibrator=None2021.11 | 0.794 | |
| PoEDirichletCalibrator=Dirichlet2021.11 | 0.794 | |
| InvRDirichletCalibrator=Dirichlet2021.11 | 0.789 | |
| InvRTempsCalibrator=Temperature Scaling2021.11 | 0.786 | |
| InvRCalibrator=None2021.11 | 0.784 | |
| CECalibrator=None2021.11 | 0.768 |