A General and Efficient Training for Transformer via Token Expansion

Open paper
Benchmark Model Rank Results
image-classification-on-cifar-10DeiT Small + ToE#24Percentage correct: 99.03
image-classification-on-cifar-100DeiT Small + ToE#37Percentage correct: 90.37
image-classification-on-imagenetDeiT Base + ToE#664Top 1 Accuracy: 81.6%

Rank counts only results with a code link.