CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR100 cluster-wise (Out)
100Cc (Data Retained)
12
CIFAR-10 (test)
94.84C-Acc (BadNet)
12
CIFAR-10 Gaussian5 Noise (test)
50.47Accuracy
12
CIFAR-10 Gaussian1 Noise (test)
60.78Accuracy
12
CIFAR-10
1.45Miscoverage (%)
12
CIFAR-10 Most-memorized 1% forget samples
0.91AUC
12
CIFAR-100 B10 Inc10
66.79Avg Performance (A-bar)
12
CIFAR-10
98.32U-LiRA Accuracy
12
CIFAR-100 non-iid Dirichlet 0.6 (test)
55.84Accuracy
12
CIFAR10 PACS (combined)
99Accuracy
12
CIFAR-10
0.388AUBC
12
CIFAR-100
44.1Error Rate
12
CIFAR10 embedding
62DR
12
CIFAR10
93.7Accuracy
12
CIFAR-10
79.72Pretrain Accuracy
12
CIFAR-10 (val)
63.64Pretrain Accuracy
12
CIFAR-10
63.64Pretrain Accuracy
12
CIFAR-10
6.74Coverage
12
CIFAR-10 (test)
94.59Baseline Accuracy
12
CIFAR-10 (synthetic)
8.99FID
12
CIFAR-10 (test)
94.58BA (DCB)
12
CIFAR-100 SSL split 2500 labels (train test)
25.52Error Rate
12
CIFAR-100 3-shot (test)
76.24Accuracy
12
CIFAR-10 (forget)
72.5AUC
12
CIFAR10
41.35Robust Accuracy
12