Image Classification on CIFAR-100 2-way Sharded (test)
63.7Test AccuracyEnsemble
Evaluation results
| Method | Links | |
|---|---|---|
| Ensemble2025.06 | 63.7 | |
| HF-Gradientalignment=Conductance2025.06 | 54.5 | |
| HF-Gradientalignment=DeepLIFT2025.06 | 54.5 | |
| KF-Gradientalignment=Conductance2025.06 | 54.4 | |
| KF-Gradientalignment=DeepLIFT2025.06 | 54.4 | |
| KF-Gradientalignment=Uniform2025.06 | 54.2 | |
| HF-Gradientalignment=Uniform2025.06 | 48.4 | |
| LPinitialization=weights of the first base model2025.06 | 45.3 | |
| KDinitialization=randomly initialized student2025.06 | 41 | |
| Individual Model (Model 1)2025.06 | 38.6 | |
| Individual Model (Model 2)2025.06 | 37.2 | |
| Transformer OTFusion wtsalignment=Conductance, fusion_data=no data used2025.06 | 4.4 | |
| Transformer OTFusion wtsalignment=DeepLIFT, fusion_data=no data used2025.06 | 4.1 | |
| Transformer OTFusion wtsalignment=Uniform, fusion_data=no data used2025.06 | 3.9 | |
| Transformer OTFusion actsalignment=Conductance, fusion_data=200 samples2025.06 | 2.3 | |
| Transformer OTFusion actsalignment=DeepLIFT, fusion_data=200 samples2025.06 | 2.3 | |
| Vanilla Averaging2025.06 | 2.2 | |
| Transformer OTFusion actsalignment=Uniform, fusion_data=200 samples2025.06 | 2.2 |