Image Classification on Image Classification Datasets (val)
94.1AccuracyGELU
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GELUModel Architecture=ResNet (BN)2026.05 | 94.1 | 43 | |
| ReLUModel Architecture=ResNet (BN)2026.05 | 93.8 | 43.6 | |
| NoisyReLUModel Architecture=ResNet (BN)2026.05 | 93.5 | 43.1 | |
| SiLUModel Architecture=ResNet (BN)2026.05 | 93.3 | 43.6 | |
| SUGARBSiLUModel Architecture=ResNet (BN)2026.05 | 92.1 | 55.5 | |
| GELUModel Architecture=MaxViT-Tiny2026.05 | 70.3 | 70.5 | |
| ReLUModel Architecture=MaxViT-Tiny2026.05 | 69.6 | 79.8 | |
| SiLUModel Architecture=MaxViT-Tiny2026.05 | 69.4 | 70.2 | |
| NoisyReLUModel Architecture=MaxViT-Tiny2026.05 | 68.5 | 79.7 | |
| ReLU2Model Architecture=MaxViT-Tiny2026.05 | 62.4 | 77.1 | |
| SUGARBSiLUModel Architecture=MaxViT-Tiny2026.05 | 51.5 | 82.9 | |
| GELUModel Architecture=MLP2026.05 | 51.1 | 63.6 | |
| SiLUModel Architecture=MLP2026.05 | 49.8 | 61 | |
| SiLUModel Architecture=MLP + RMSNorm2026.05 | 49.4 | 62.3 | |
| NoisyReLUModel Architecture=MLP2026.05 | 49.1 | 70.4 | |
| SUGARBSiLUModel Architecture=MLP + RMSNorm2026.05 | 48.6 | 70.8 | |
| GELUModel Architecture=MLP + RMSNorm2026.05 | 48.4 | 64.5 | |
| NoisyReLUModel Architecture=MLP + RMSNorm2026.05 | 48.4 | 66.8 | |
| ReLU2Model Architecture=MLP + RMSNorm2026.05 | 48 | 65.3 | |
| ReLUModel Architecture=MLP2026.05 | 46.6 | 70.8 | |
| ReLU2Model Architecture=MLP + LayerNorm2026.05 | 45.3 | 62.5 | |
| ReLUModel Architecture=MLP + RMSNorm2026.05 | 44.7 | 68.4 | |
| SiLUModel Architecture=MLP + LayerNorm2026.05 | 43.9 | 65.4 | |
| GELUModel Architecture=MLP + LayerNorm2026.05 | 42.3 | 67.2 | |
| NoisyReLUModel Architecture=MLP + LayerNorm2026.05 | 41.8 | 66.4 | |
| SUGARBSiLUModel Architecture=MLP + LayerNorm2026.05 | 41 | 58.5 | |
| ReLUModel Architecture=MLP + LayerNorm2026.05 | 40.9 | 68.6 | |
| SUGARBSiLUModel Architecture=MLP2026.05 | 40.8 | 72.5 | |
| ReLU2Model Architecture=MLP2026.05 | 10 | 8 | |
| ReLU2Model Architecture=ResNet (BN)2026.05 | 10 | 0 |