CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR100 32/255 noise (test)
49.38Natural Accuracy
10
CIFAR100 16/255 noise (test)
50.76Natural Accuracy
10
CIFAR100 8/255 noise (test)
60.29Natural Accuracy
10
CIFAR10 32/255 noise (test)
70.84Natural Accuracy
10
CIFAR10 16/255 noise (test)
82.98Natural Accuracy
10
CIFAR10 12/255 noise (test)
81.09Natural Accuracy
10
CIFAR10 (test)
46.95Robust Acc (8/255)
10
CIFAR-10 1 class
99.1AUROC
10
CIFAR-100
75.36MiniGPT-4 Score
10
CIFAR-10 (test)
98.33MiniGPT-4 Accuracy
10
CIFAR-100 (test)
58.5Avg Clean Acc
10
CIFAR100
77.38Accuracy
10
CIFAR10
93.49Accuracy
10
CIFAR100-I2C Transfer from ImageNet100 (test)
82.3Accuracy
10
CIFAR-10 beta=0.3 (test)
10.85MAE
10
CIFAR-5
22.67Test Error (mean)
10
CIFAR-10 (test)
82.1Clean Accuracy
10
CIFAR-100 Hue-shifted (test)
71.33Accuracy
10
CIFAR-10 Hue-shifted (test)
93.8Accuracy
10
CIFAR-100 400 labels (test)
78.8Top-1 Acc
10
CIFAR-100 200 labels (test)
74.17Top-1 Accuracy
10
CIFAR-10 original (test)
2.65BPD
10
CIFAR100 long-tailed (train)
5.81FID
10
CIFAR-FS
96Mean Accuracy
10
CIFAR-FS
90.3Mean Accuracy
10