Image Classification on Tiny-ImageNet (Sharded 2-way split)
44.5AccuracyEnsemble
Evaluation Results
| Method | Links | |
|---|---|---|
| Ensemblesplit=2-way split2025.06 | 44.5 | |
| HF-Gradient DeepLIFTsplit=2-way split, alignment=DeepLIFT2025.06 | 32.5 | |
| HF-Gradient Conductancesplit=2-way split, alignment=Conductance2025.06 | 32.4 | |
| KF-Gradient Conductancesplit=2-way split, alignment=Conductance2025.06 | 32.4 | |
| KF-Gradient Uniformsplit=2-way split, alignment=Uniform2025.06 | 32.3 | |
| KF-Gradient DeepLIFTsplit=2-way split, alignment=DeepLIFT2025.06 | 31.5 | |
| LPsplit=2-way split2025.06 | 30.8 | |
| HF-Gradient Uniformsplit=2-way split, alignment=Uniform2025.06 | 30.6 | |
| Individual Model 0split=2-way split2025.06 | 28.3 | |
| Individual Model 1split=2-way split2025.06 | 27.5 | |
| KDsplit=2-way split2025.06 | 11.3 | |
| Transformer OTFusion wts Uniformsplit=2-way split, variant=weights-based2025.06 | 4 | |
| Transformer OTFusion acts Conductancesplit=2-way split, variant=activations-based2025.06 | 2.7 | |
| Transformer OTFusion acts Uniformsplit=2-way split, variant=activations-based2025.06 | 2.6 | |
| Transformer OTFusion acts DeepLIFTsplit=2-way split, variant=activations-based2025.06 | 2.6 | |
| Transformer OTFusion wts Conductancesplit=2-way split, variant=weights-based2025.06 | 2.3 | |
| Transformer OTFusion wts DeepLIFTsplit=2-way split, variant=weights-based2025.06 | 2.1 | |
| Vanilla Averagingsplit=2-way split2025.06 | 0.9 |