Sharpness-Aware Minimization for Efficiently Improving Generalization

Open paper
Benchmark Model Rank Results
fine-grained-image-classification-on-birdsnapEffNet-L2 (SAM)#1Accuracy: 90.07%
fine-grained-image-classification-on-fgvcEffNet-L2 (SAM)#42Top-1 Error Rate: 4.82
fine-grained-image-classification-on-food-101EffNet-L2 (SAM)#2Accuracy: 96.18
fine-grained-image-classification-on-oxford-2EffNet-L2 (SAM)#1Accuracy: 97.10Top-1 Error Rate: 2.90%
fine-grained-image-classification-on-stanfordEffNet-L2 (SAM)#7Accuracy: 95.96%
image-classification-on-cifar-100EffNet-L2 (SAM)#1Percentage correct: 96.08
image-classification-on-cifar-100PyramidNet (SAM)#27Percentage correct: 89.7
image-classification-on-cifar-100CNN39#184Percentage correct: 42.64
image-classification-on-cifar-100CNN36#185Percentage correct: 36.07
image-classification-on-flowers-102EffNet-L2 (SAM)#6Accuracy: 99.65%
image-classification-on-imagenetEfficientNet-L2-475 (SAM)#34Top 1 Accuracy: 88.61%Number of params: 480M
image-classification-on-imagenetResNet-152 (SAM)#598Top 1 Accuracy: 81.6%

Rank counts only results with a code link.