CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR100 Fewshot (test)
45.1Accuracy (1-shot)
6
CIFAR-100
75.93Accuracy (10% Noise)
6
CIFAR-10 + CIFAR-100 open-set noise (test)
79.55Accuracy
6
CIFAR100
66.71-shot Acc
6
CIFAR-100 32x32x3 (test)
98AUPR-Out (C0)
6
CIFAR-100 unknown attack 1.0 (test)
99.86AUROC (FGSM seen)
6
CIFAR-10 unknown attack 1.0 (test)
99.94AUROC (FGSM seen)
6
CIFAR10 (test)
72.2Accuracy
6
CIFAR-10 no data augmentation (test)
8.27Test Error Rate
6
CIFAR-10 (Pat-2)
224Communication Rounds (Acc@90)
6
CIFAR-10 (test)
85Success Count
6
CIFAR100 (train+test)
36.2NMI
6
CIFAR-100 (test)
42.19Mf Score
5
CIFAR-10
11.77Inference Time (s)
5
CIFAR-10 (test)
93.11Standard Accuracy
5
CIFAR-10-C Severity Level 3 (test)
77.68Mean Test Accuracy
5
CIFAR-100 Clean standard (test)
16.41AURC
5
CIFAR-100 (test)
42.4Accuracy
5
CIFAR-10
23.43ASR
5
CIFAR-10 & CIFAR-100 cross-data
0.79R^2
5
CIFAR10 ResNet18 (test)
94.7Utility Before Unlearning
5
CIFAR-100
60.56Accuracy (SVHN, S)
5
CIFAR-10
91.5SVHN AccS
5
CIFAR-10 and CIFAR-100 grid of 1,152 models (test)
0.902Spearman ρ
5
CIFAR-10 (test)
74.72Top-1 Accuracy
5