Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding

Open paper
Benchmark Model Rank Results
image-classification-on-cifar-10Transformer local-attention (NesT-B)#112Percentage correct: 97.2
image-classification-on-cifar-100Transformer local-attention (NesT-B)#131Percentage correct: 82.56
image-classification-on-imagenetTransformer local-attention (NesT-B)#415Top 1 Accuracy: 83.8%Number of params: 68MGFLOPs: 17.9
image-classification-on-imagenetTransformer local-attention (NesT-S)#473Top 1 Accuracy: 83.3%Number of params: 38MGFLOPs: 10.4
image-classification-on-imagenetTransformer local-attention (NesT-T)#669Top 1 Accuracy: 81.5%Number of params: 17MGFLOPs: 5.8

Rank counts only results with a code link.