Plausibility Analysis on CIFAR-100 original (val)
1RankAttention (SMaT)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Attention (SMaT)Backbone=Vision Transformer (ViT), Pre-training=ImageNet-21k, Resolution=224 x 2242022.04 | 1 | — | |
| Attention (all layers)Backbone=Vision Transformer (ViT), Pre-training=ImageNet-21k, Resolution=224 x 2242022.04 | 2 | — |