CIFAR10
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR10 unlabeled 1 (train)
91.42Accuracy
5
CIFAR10 epsilon = 4/255 (test)
35.26Clean Error Rate
5
CIFAR10 LT
17.74FID
5
CIFAR10 MCLs setting (test)
82.33Natural Accuracy
5
CIFAR10 IDN 0.9 (test)
11.01Accuracy
5
CIFAR10 IDN 0.7 (test)
47.23Accuracy
5
CIFAR10 80 labels (0.16%) (test)
95.2Accuracy
5
CIFAR10 Densenet121 (test)
84Accuracy
5
Cifar10 150K iterations (test)
8.69Error Rate
5
CIFAR10 50% label noise (test)
80.7Test Accuracy
5
CIFAR10 25% label noise (test)
82.9Test Accuracy
5
CIFAR10 GNNBenchmark (test)
74.754Accuracy
5
CIFAR10
61Concept Coverage
4
CIFAR10
86.13Accuracy
4
CIFAR10 5 Tasks
69.76Accuracy
4
CIFAR10 (test)
81.27Accuracy
4
CIFAR10 compatible-shift
81.95Old-task Accuracy
4
CIFAR10 10% label noise (test)
4.75Test Error Rate
4
CIFAR10 0
61.91FID
4
CIFAR10.1 OOD (test)
91.22Accuracy
4
CIFAR10-LT vs SVHN
98.98AUROC
4
CIFAR10 Unlearning normal samples (De)
68.28Accuracy
4
CIFAR10 Unlearning normal samples (test)
75.21Accuracy
4
CIFAR10 Unlearning backdoored samples (De)
8.11Accuracy
4
CIFAR10 Unlearning backdoored samples (test)
75.83Accuracy
4