Image Classification on CIFAR-100 (test) (Accuracy and Efficiency)
86.94AccuracyAdaptive-α
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Adaptive-αBackbone=ViT-B-MAE, Training Protocol=Finetuned from ImageNet2026.04 | 86.94 | 55.8 | 0.13 | 4.6 | |
| BaselineBackbone=ViT-B-MAE, Training Protocol=Finetuned from ImageNet2026.04 | 86.81 | 200 | — | — | |
| Adaptive-TBackbone=ViT-B-MAE, Training Protocol=Finetuned from ImageNet2026.04 | 85.55 | 43.5 | 0.13 | 4.6 | |
| Adaptive-TBackbone=EfficientNet-B0, Training Protocol=Finetuned from ImageNet2026.04 | 84.11 | 20.8 | 0.8 | 9.62 | |
| Adaptive-αBackbone=EfficientNet-B0, Training Protocol=Finetuned from ImageNet2026.04 | 83.98 | 26.1 | 0.8 | 9.62 | |
| BaselineBackbone=EfficientNet-B0, Training Protocol=Finetuned from ImageNet2026.04 | 83.31 | 200 | — | — | |
| Adaptive-αBackbone=EfficientNet-B0, Training Protocol=From Scratch2026.04 | 66.51 | 23.94 | 0.19 | 8.36 | |
| BaselineBackbone=EfficientNet-B0, Training Protocol=From Scratch2026.04 | 66.32 | 200 | — | — | |
| Adaptive-TBackbone=EfficientNet-B0, Training Protocol=From Scratch2026.04 | 63.67 | 21.13 | 0.19 | 8.36 | |
| Adaptive-αBackbone=MobileNet-V2, Training Protocol=From Scratch2026.04 | 63.14 | 25.61 | 1.89 | 7.84 | |
| Adaptive-TBackbone=MobileNet-V2, Training Protocol=From Scratch2026.04 | 62.99 | 22.62 | 1.89 | 7.84 | |
| BaselineBackbone=MobileNet-V2, Training Protocol=From Scratch2026.04 | 61.25 | 200 | — | — |