Image Classification on Tiny-ImageNet (Epochs to convergence, Improvement (%))
35.6Epochs to ConvergenceCT-AGD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CT-AGDModel=DeiT-12, Target Accuracy (%)=30.42026.05 | 35.6 | 61 | |
| CT-AGDModel=WRN-28-4, Target Accuracy (%)=55.52026.05 | 54.5 | 32.2 | |
| SGDModel=WRN-28-4, Target Accuracy (%)=55.52026.05 | 80.4 | — | |
| AdamModel=WRN-28-4, Target Accuracy (%)=55.52026.05 | 87.8 | — | |
| SGDModel=DeiT-12, Target Accuracy (%)=30.42026.05 | 91.2 | — | |
| AdamModel=DeiT-12, Target Accuracy (%)=30.42026.05 | 96.2 | — | |
| SGDModel=ResNet-20, Target Accuracy (%)=44.32026.05 | 113 | — | |
| CT-AGDModel=ResNet-20, Target Accuracy (%)=44.32026.05 | 116.5 | -3.09 | |
| AdamModel=ResNet-20, Target Accuracy (%)=44.32026.05 | 145.2 | — |