Image Classification on CIFAR-100 (test) (Accuracy Profile by Training Epoch)
68Top-1 Accuracy (Epoch 50)Lion
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LionModel=ResNet-502026.06 | 68 | 70.39 | 72.99 | |
| LPSGDMModel=ResNet-502026.06 | 67.47 | 71.45 | 76.78 | |
| AdamWModel=ResNet-502026.06 | 65.84 | 67.55 | 74.91 | |
| LionModel=ResNet-182026.06 | 65.39 | 67.64 | 73.07 | |
| SGD w/ MomentumModel=ResNet-502026.06 | 63.34 | 68.08 | 69.7 | |
| LionModel=VGG-112026.06 | 55.92 | 56.27 | 58.74 | |
| LPSGDMModel=VGG-112026.06 | 55.48 | 57.78 | 60.46 | |
| SGD w/ MomentumModel=VGG-112026.06 | 54.21 | 58.33 | 60.16 | |
| AdamWModel=VGG-112026.06 | 51.01 | 55.11 | 59.34 | |
| LPSGDMModel=ResNet-182026.06 | 48.03 | 54.31 | 77.98 | |
| AdamWModel=ResNet-182026.06 | 47.92 | 53.85 | 74.23 | |
| SGD w/ MomentumModel=ResNet-182026.06 | 44.12 | 53.14 | 76.86 |