Multi-class classification on CIFAR-100 (test) (Top1, AUC, ECE, SCE, ACE, smECE)
56.83Top-1 AccuracyRefCal
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RefCalBackbone=EfficientViT-M1, Reference Baseline=LS2026.05 | 56.83 | 96.2 | 8.32 | 25 | 19 | 7.9 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=MbLS2026.05 | 56.82 | 96.2 | 8.5 | 26 | 19 | 8.07 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=NLL (CE)2026.05 | 56.78 | 96 | 9.67 | 28 | 19 | 9.17 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=CE+TS2026.05 | 56.78 | 96.3 | 7.82 | 25 | 26 | 7.55 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=CRL2026.05 | 56.75 | 96.1 | 9.24 | 27 | 18 | 8.79 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=MMCE2026.05 | 56.74 | 96 | 9.61 | 28 | 19 | 9.11 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=FL+ MDCA2026.05 | 56.69 | 96.6 | 5.9 | 24 | 20 | 5.59 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=LogitNorm2026.05 | 55.93 | 95.3 | 25.93 | 60 | 34 | 22.92 | |
| RefCalBackbone=EfficientViT-M1, Reference Baseline=AdaFocal2026.05 | 53.1 | 93.9 | 23.3 | 57 | 43 | 21.42 | |
| AdaFocalBackbone=EfficientViT-M12026.05 | 52.39 | 96.63 | 14.35 | 43 | 29 | 14.36 | |
| MbLSBackbone=EfficientViT-M12026.05 | 49.9 | 93.85 | 8.78 | 33 | 25 | 8.78 | |
| CRLBackbone=EfficientViT-M12026.05 | 49.07 | 95.11 | 28.53 | 69 | 41 | 26.45 | |
| LSBackbone=EfficientViT-M12026.05 | 48.97 | 92.79 | 4.11 | 30 | 25 | 4.06 | |
| MMCEBackbone=EfficientViT-M12026.05 | 48.7 | 95.11 | 28.82 | 70 | 42 | 26.61 | |
| NLL (CE)Backbone=EfficientViT-M12026.05 | 47.77 | 95.07 | 29.53 | 71 | 42 | 27.13 | |
| CE+TSBackbone=EfficientViT-M12026.05 | 47.77 | 95.4 | 15.25 | 43 | 35 | 14.99 | |
| FL+ MDCABackbone=EfficientViT-M12026.05 | 47.25 | 95.25 | 17.76 | 51 | 37 | 17.69 | |
| LogitNormBackbone=EfficientViT-M12026.05 | 44.61 | 94.68 | 16.38 | 47 | 35 | 16.36 |