CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10
74.52Accuracy
30
CIFAR100
77.5Accuracy
30
CIFAR10
94.37Accuracy
30
CIFAR-100
56.2Accuracy
30
CIFAR-100 20 tasks (test)
69.9Average Novelty
30
CIFAR-10 (test)
93.62Standard Accuracy
30
CIFAR-10 @1000 (seen classes)
97.3Seen@Seen mAP
30
CIFAR-FS
90.26Mean Accuracy
30
CIFAR-100 8% utility loss, ϵ = 10, δ = 10⁻⁵ (test)
0.64MLS (TPR@0.1% FPR) Pre-Hardening
30
CIFAR-10 10% utility loss, ϵ = 8, δ = 10⁻⁵ (test)
45.44MLS (Pre-Hardening, TPR@0.1% FPR)
30
CIFAR-100 (test)
69.98Accuracy
30
CIFAR-10 (test)
98.52Accuracy (Retain)
30
CIFAR-10 Imbalance Factor 10 Long-Tailed (test)
70.5Accuracy
30
CIFAR10 6 closed, 4 open classes 1.0
0.997AUROC
30
CIFAR-10 Dirichlet partition (test)
95Communication Rounds
30
CIFAR-100 10 images/class (test)
46.2Accuracy
30
CIFAR-10 corrupted (test)
89.4Acc
30
CIFAR-10 100% data
10.38IS
30
CIFAR-10
99.2FGSM Robust Accuracy
30
CIFAR-100 90% symmetric label noise
58.56Accuracy
30
CIFAR-100 80% symmetric label noise
58.88Accuracy
30
CIFAR10 (test)
3.12Bits Per Dimension (BPD)
30
CIFAR-100 1.0 (test)
77.78Accuracy
30
CIFAR-10 18 (test)
95.69Accuracy (ACC)
29
CIFAR-10 (test)
95.4Accuracy
29