Pay Attention to Attention Distribution: A New Local Lipschitz Bound for Transformers
Open paper
Benchmark
Model
Rank
Results
image-classification-on-cifar-10
JaSMin (ViT-B, k=0, lambda=1e-2, mean)
–
Percentage correct: 98.55
image-classification-on-cifar-100
JaSMin (ViT-B, k=10, lambda=1e-3, max)
–
Percentage correct: 90.94
Rank counts only results with a code link.