Image Classification on CIFAR-10 80% flips Standard
0.5658AccuracyREFINE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| REFINEBackbone=CNNs2025.05 | 0.5658 | 0.9067 | 0.5655 | 0.369 | |
| NoTransBackbone=CNNs2025.05 | 0.5657 | 0.9057 | 0.5622 | 0.336 | |
| DistillBackbone=CNNs2025.05 | 0.5351 | 0.8982 | 0.5269 | 0.268 | |
| LoRABackbone=CNNs2025.05 | 0.2292 | 0.7202 | 0.1911 | 0.0076 | |
| DANN-GateBackbone=CNNs2025.05 | 0.2083 | 0.7097 | 0.1341 | 0 | |
| LinearProbeBackbone=CNNs2025.05 | 0.1946 | 0.6895 | 0.1177 | 0 | |
| AdapterBackbone=CNNs2025.05 | 0.1849 | 0.6906 | 0.1219 | 0 |