Selective Classification on ImageNet V2 Original (test)
30.4Metric AKNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| KNNBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 30.4 | 1.11 | |
| MDSBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 21.6 | 0.736 | |
| SIRCBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 19.6 | 0.652 | |
| EnergyBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 17.6 | 0.567 | |
| Δ-KNNBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 16.2 | 0.509 | |
| Δ-MDSBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 13.7 | 0.399 | |
| MaxLogitBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 13.5 | 0.394 | |
| MSPBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 10.32 | 0.257 | |
| RLogBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 9.14 | 0.207 | |
| Δ-KNN-RLogBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 9.02 | 0.202 | |
| Δ-MDS-RLogBackbone=ResNet-50, Training Dataset=ImageNet-1K, Pre-training Source=Official PyTorch Repository2025.05 | 8.86 | 0.195 |