Image Classification on CelebA WILDS (test)
95.6I.I.D. AccuracyERM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ERMBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 95.6 | 47.2 | |
| uLABias labels used for training (TR.)=false, Bias labels used for validation (VAL)=true2025.12 | 93.9 | 86.5 | |
| ERMModel size=B/162026.03 | 93.9 | 34.2 | |
| ERMModel size=L/162026.03 | 93.9 | 35.9 | |
| GroupDROBias labels used for training (TR.)=true, Bias labels used for validation (VAL)=true2025.12 | 92.9 | 88.9 | |
| σ−stitchingModel size=B/162026.03 | 92.6 | 46.1 | |
| RWYModel size=B/162026.03 | 92.3 | 42.6 | |
| σ−stitchingModel size=L/162026.03 | 92.2 | 41.9 | |
| RandARModel size=L/162026.03 | 92 | 63.9 | |
| RWYModel size=L/162026.03 | 91.4 | 45.3 | |
| MTBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 91.3 | 78 | |
| AFRBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 91.3 | 82 | |
| DFRBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=true2025.12 | 91.3 | 88.3 | |
| RandARModel size=B/162026.03 | 90.8 | 60 | |
| DiTModel size=L/162026.03 | 90.4 | 59.6 | |
| CNCBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 89.9 | 88.8 | |
| DiTModel size=B/162026.03 | 89.9 | 60.1 | |
| NSFBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 88.7 | 84.27 | |
| JTTBias labels used for training (TR.)=false, Bias labels used for validation (VAL)=false2025.12 | 88 | 81.1 |