Image Classification on ImageNet ILSVRC 2012 (val) with Calibration Metrics
79.5Top-1 AccuracyHet. Ensemble 4x
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Het. Ensemble 4xBackbone=ResNet-50, Ensemble size=4, Temperature=1.5, Covariance rank=15, Training epochs=1802021.05 | 79.5 | 0.79 | 0.015 | |
| Hom. Ensemble 4xBackbone=ResNet-50, Ensemble size=4, Training epochs=902021.05 | 77.5 | 0.877 | 0.0305 | |
| Het. Single ModelBackbone=ResNet-50, Ensemble size=1, Temperature=1.5, Covariance rank=15, Training epochs=1802021.05 | 77.5 | 0.898 | 0.033 | |
| Hom. Single ModelBackbone=ResNet-50, Ensemble size=1, Training epochs=902021.05 | 76.1 | 0.943 | 0.0392 |