CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR10 0.1
87.02Accuracy
13
CIFAR-100
79.04ID Accuracy
13
CIFAR-10 LT
91.24Acc (Ratio 10)
13
CIFAR-10 (test)
97.9Accuracy (Cluster, All)
13
CIFAR100-20 unlabelled (train)
92.4Clustering Accuracy
13
CIFAR-10 20% asymmetric noise
95.4Accuracy
13
CIFAR-100-LT ($
ho=100$) (test)
48.9Accuracy
13
CIFAR-10 (test)
88.71Clean Accuracy
13
CIFAR 10 latent dimension 32 (test)
93.53FID
13
CIFAR-10 vs CIFAR-100
0.967ROCAUC
13
CIFAR-10
0.949Weighted Kendall's Tau
13
CIFAR10
0.991AUROC
13
CIFAR-10 (32 x 32) Conditional (test)
2.25FID
13
CIFAR-100-LT gamma=20, beta=40% (test)
57.6Overall Accuracy
13
CIFAR-10
96.5Accuracy
13
CIFAR-100 4k labels
28.61Error Rate
13
CIFAR-10 (val)
95.6Acc (20% noise)
13
CIFAR-10 (test)
1.54C10+ Error Rate
13
CIFAR-100 vs LSUN (test)
16.16FPR @ FNR=0.05
13
CIFAR10
3.12BPD
13
CIFAR-10 (test)
64Validity Count
12
CIFAR-10 (test)
24.51Relative Top-1 Accuracy Change (pp)
12
CIFAR-100 (20 / 80)
64.55Balanced Accuracy
12
CIFAR10
58Retention Accuracy (R-A)
12
CIFAR-100
44.2ACC
12