CIFAR-10
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 Target Model WRN-34-10
41.9PGD Attack Success Rate (L-inf)
24
CIFAR-10-C severity 5 (test)
79.26Gaussian Noise Accuracy (Severity 5)
24
CIFAR-10 ID Average OOD
93.67AUROC
24
CIFAR-10 Split 5 sequential tasks (test)
0Final Forgetting (FF)
24
CIFAR-10
862,998Noise Scale (L2 norm)
24
S-CIFAR-10 (last)
0.983TIL Last Top-1 Acc
24
CIFAR-10 standard (test)
2.34GFLOPS
24
CIFAR-10 Split
64.83Final Avg Accuracy
24
CIFAR-10 Split
0.22Forgetting Measure
24
CIFAR-10 1.0 (val)
80.3Accuracy
24
CIFAR-10 (test)
86.84Robust Accuracy
24
CIFAR-10 (test)
2.46Test Error Rate
22
CIFAR-10 LSUN-Resize semantic OOD + CIFAR-10-C covariate OOD (test)
0.02FPR
22
CIFAR-10-C (target clients)
75.37Accuracy
22
CIFAR-10 Split (test)
92.2Average Accuracy (AT)
22
Split CIFAR-10 5 disjoint subsets
92.53ACC
22
CIFAR-10 standard (test)
95.94Top-1 Accuracy
22
CIFAR-10 standard (test)
2.83Top-1 Error
22
CIFAR-10N Worst noise
79.2Test Accuracy
21
CIFAR-10 redundant expert suite (val)
91.9System Accuracy
21
CIFAR-10 3 classes (forget)
94Accuracy
21
CIFAR-10 30% symmetric noise (test)
82.26Accuracy
21
CIFAR-10 Sequential (test)
52.3Accuracy
21
CIFAR-10
87.87ATPF
21
CIFAR-10 30% Minority (test)
87.75Majority Accuracy
20