Image Classification on CIFAR100 (80:20) Corrupted (test)
66.8AccuracySTAMP + Hi-Vec
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| STAMP + Hi-VecBackbone=ResNet-182025.08 | 66.8 | 72.1 | 69.3 | |
| STAMPBackbone=ResNet-182025.08 | 66 | 71.2 | 68.5 | |
| SAR + Hi-VecBackbone=ResNet-182025.08 | 63.8 | 69.7 | 66.6 | |
| SARBackbone=ResNet-182025.08 | 63.4 | 69.1 | 66.1 | |
| SoTTABackbone=ResNet-182025.08 | 61.4 | 68.2 | 64.6 | |
| EATABackbone=ResNet-182025.08 | 61.3 | 67.4 | 64.2 | |
| Tent + Hi-VecBackbone=ResNet-182025.08 | 60.6 | 68.2 | 64.1 | |
| TentBackbone=ResNet-182025.08 | 60.4 | 65.6 | 62.8 | |
| CoTTABackbone=ResNet-182025.08 | 57.1 | 66.3 | 61.3 | |
| OWTTTBackbone=ResNet-182025.08 | 56.2 | 66 | 60.6 | |
| BN StatsBackbone=ResNet-182025.08 | 55.5 | 66.2 | 60.2 | |
| RoTTABackbone=ResNet-182025.08 | 50.6 | 63.7 | 56.2 | |
| MRLBackbone=ResNet-182025.08 | 43.4 | 62.2 | 50.3 | |
| SourceBackbone=ResNet-182025.08 | 37.1 | 60.6 | 44.2 |