Image Classification on CIFAR-100 (Accuracy and Calibration)
92.17AccuracyConvNeXt-Large
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ConvNeXt-LargeRole=Teacher2026.06 | 92.17 | 91.89 | 0.0049 | |
| ViT-B/16Role=Teacher2026.06 | 91.51 | 92.72 | 0.0174 | |
| MobileNetV2 (KD + MixUp)Architecture=MobileNetV2, Teacher Model=ConvNeXt-Large, Training Protocol=KD + MixUp2026.06 | 86.3 | 87.52 | 0.0203 | |
| MobileNetV2 (KD + MixUp)Architecture=MobileNetV2, Teacher Model=ViT-B/16, Training Protocol=KD + MixUp2026.06 | 85.8 | 88.76 | 0.0315 | |
| MobileNetV2 (KD + MixUp)Architecture=MobileNetV2, Teacher Model=ResNet152V2, Training Protocol=KD + MixUp2026.06 | 84.1 | 88.87 | 0.0496 | |
| ResNet152V2Role=Teacher2026.06 | 82.57 | 88.73 | 0.0619 | |
| ConvNeXt-TinyRole=Teacher2026.06 | 81.96 | 79.08 | 0.0288 | |
| MobileNetV2 (KD + MixUp)Architecture=MobileNetV2, Teacher Model=ConvNeXt-Tiny, Training Protocol=KD + MixUp2026.06 | 81.1 | 77.48 | 0.0364 | |
| MobileNetV2 BaselineArchitecture=MobileNetV2, Training Protocol=Baseline2026.06 | 81.05 | 95.27 | 0.1421 |