Learning Spatial Decay for Vision Transformers

Benchmark Model Rank Results
image-classification-on-imagenetSDT-H-B–Top 1 Accuracy: 85.1%GFLOPs: 10.8
image-classification-on-imagenetSDT-H-S–Top 1 Accuracy: 84.2%GFLOPs: 4.8
image-classification-on-imagenetSDT-H-T–Top 1 Accuracy: 82.7%GFLOPs: 2.7