Image Classification on CIFAR-10 (val) (Accuracy)
92.15AccuracySignMuon
Evaluation results
| Method | Links | |
|---|---|---|
| SignMuonBackbone=ResNet-50, BS=512, LR=0.001, NS=1, WD=0.02026.05 | 92.15 | |
| Dist-SignMuonBackbone=ResNet-50, BS=128, LR=0.001, NS=1, WD=0.02026.05 | 92.02 | |
| Dist-SignMuonBackbone=ResNet-50, BS=512, LR=0.001, NS=1, WD=0.02026.05 | 90.56 | |
| AdamBackbone=ResNet-50, BS=512, LR=0.0012026.05 | 89.71 | |
| Dist-MuonBackbone=ResNet-50, BS=128, LR=0.01, NS=1, WD=0.02026.05 | 89.09 | |
| MuonBackbone=ResNet-50, BS=512, LR=0.1, NS=1, WD=0.02026.05 | 89.02 | |
| Dist-AdamBackbone=ResNet-50, BS=128, LR=0.0012026.05 | 88.9 | |
| SignAdamBackbone=ResNet-50, BS=512, LR=0.0012026.05 | 88.24 | |
| Dist-SignAdamBackbone=ResNet-50, BS=128, LR=0.00012026.05 | 87.72 | |
| SignSGDBackbone=ResNet-50, BS=512, LR=0.0012026.05 | 87.44 |