ImageNet
Benchmarks
TaskDatasetSOTA resultTrendResults
ImageNet Robustness Suite
93.96Top-1 Accuracy (ImageNet-A)
89
Tiny-ImageNet
56TER
88
ImageNet 64x64 (val)
1.01FID
87
ImageNet100 (test)
86.74Top-1 Acc
87
Mini-Imagenet 5-way 5-shot
98Accuracy
87
tieredImagenet (test)
98.1Accuracy
86
ImageNet-1K 256x256 (val)
1.28gFID
86
ImageNet-54 (test)
99.2AUC
85
TinyImageNet
70.02Avg Incremental Accuracy
83
ImageNet 64x64 (train val)
0.97FID
83
ImageNet-R (val)
96.5Accuracy
82
ImageNet 1000 images (val)
86.4Robust Accuracy
82
Mini-ImageNet 1-shot 5-way (test)
85.59Accuracy
82
ImageNet class-conditional 256x256 (test val)
1.55FID
81
ImageNet-100 1.0 (test)
0.47JSD Correlation to Accuracy Diff
80
ImageNet (val)
59.06Accuracy
80
ImageNet
7.05Interpretation Error
80
ImageNet 21k (val)
0.694Median L2 Distance
80
ImageNet-1K 1.0 (val)
79.3k-NN Accuracy
80
ImageNet-1k (val)
65.2Top-1 Loc Acc
80
ImageNet 1k (val)
1.13FID
79
ImageNet-1k (val)
0.334EHR
79
ImageNet-1k
82.92Top-1 Accuracy
78
ImageNet-1K
83.6Top-1 Accuracy
78
ImageNet 1k (val)
34.71Insertion Score
78