Image Classification on CIFAR-10 (test) (k-NN Accuracy)
39.51Test Accuracy (k=2)UWA
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| UWABackbone=ResNet-18, Pre-training=trained from scratch, Number of seeds=32025.09 | 39.51 | 59.37 | 64.71 | 68.96 | 71.23 | 73.16 | 80.84 | |
| sUWABackbone=ResNet-18, Pre-training=trained from scratch, Number of seeds=32025.09 | 38.33 | 60.96 | 67.45 | 71.38 | 73.94 | 74.98 | — | |
| AVGBackbone=ResNet-18, Pre-training=trained from scratch, Number of seeds=32025.09 | 21.7 | 41.88 | 56.79 | 65.33 | 72.55 | 75.42 | — |