Classification on FairFace
80.8Mean AccuracyF-Layer
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| F-LayerBackbone=Swin-T2026.05 | 80.8 | — | — | |
| F-LayerBackbone=ViT-B/162026.05 | 77.5 | — | — | |
| ProjectionBackbone=Swin-T2026.05 | 76.5 | — | — | |
| F-LayerBackbone=ResNet-182026.05 | 76.2 | — | — | |
| F-LayerBackbone=DenseNet-1212026.05 | 76.2 | — | — | |
| ProjectionBackbone=ViT-B/162026.05 | 75.4 | — | — | |
| ProjectionBackbone=ResNet-182026.05 | 74.8 | — | — | |
| ProjectionBackbone=DenseNet-1212026.05 | 73.9 | — | — | |
| F-LayerBackbone=CustomCNN2026.05 | 67.3 | — | — | |
| ProjectionBackbone=CustomCNN2026.05 | 67 | — | — | |
| PenaltyBackbone=Swin-T2026.05 | 55.1 | — | — | |
| Strict PenaltyBackbone=Swin-T2026.05 | 55.1 | — | — | |
| PenaltyBackbone=DenseNet-1212026.05 | 55.1 | — | — | |
| Strict PenaltyBackbone=DenseNet-1212026.05 | 55.1 | — | — | |
| PenaltyBackbone=ResNet-182026.05 | 55 | — | — | |
| Strict PenaltyBackbone=ResNet-182026.05 | 55 | — | — | |
| PenaltyBackbone=ViT-B/162026.05 | 55 | — | — | |
| Strict PenaltyBackbone=ViT-B/162026.05 | 55 | — | — | |
| PenaltyBackbone=CustomCNN2026.05 | 55 | — | — | |
| Strict PenaltyBackbone=CustomCNN2026.05 | 55 | — | — | |
| StableRep+training dataset=cc12m, pre-train data=Syn, zero-shot=true2023.06 | 37.2 | 74.9 | 10 | |
| CLIPtraining dataset=cc12m, pre-train data=Syn, zero-shot=true2023.06 | 30.4 | 64 | 3.1 | |
| CLIPtraining dataset=cc12m, pre-train data=Real, zero-shot=true2023.06 | 28.2 | 60.2 | 0.3 | |
| StableRep+training dataset=redcaps, pre-train data=Syn, zero-shot=true2023.06 | 27.3 | 64.4 | 2.1 | |
| CLIPtraining dataset=redcaps, pre-train data=Syn, zero-shot=true2023.06 | 22.3 | 52.4 | 1 | |
| CLIPtraining dataset=redcaps, pre-train data=Real, zero-shot=true2023.06 | 9.3 | 31.1 | 0.4 |