ImageNet
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Restricted ImageNet (test)
87.5Clean Accuracy
6
ImageNet 1K 256x256 (val)
35.65PSNR
6
ImageNet 1K 256x256 (val)
26.55PSNR
6
ImageNet 1K noise sigma=0.05 256x256
25.46PSNR
6
ImageNet 1K noise sigma=0.05 256x256
25.22PSNR
6
ImageNet
99.9Attack Success Rate (Dense-121)
6
ImageNet 1k (test)
90.4Model Performance (IncRes-v2)
6
ImageNet
73.7AT Accuracy (L2)
6
mini-ImageNet 10000 labels (train)
39.43Top-1 Error
6
mini-ImageNet 4000 labels (train)
46.32Top-1 Error Rate
6
ImageNet-100 B50, 25 steps (test)
0.812Avg Accuracy
6
ImageNet (test)
77.5Clean Accuracy
6
ImageNet 10% labels (val)
84.3Top-1 Acc
6
ImageNet 1% labels (val)
80Top-1 Acc
6
ImageNet 1k V2 R A (test)
86.3Top-1 Acc (Test)
6
ImageNet-C OOD-domains
1.66MAE
6
Tiny-ImageNet Sequential
13.1Forgetting
6
Tiny-ImageNet Sequential
32.45Forgetting
6
ImageNet-100 (test)
28.7Intra-FID
6
ImageNet-291 Bird -> Car (test)
13.93FID
6
ImageNet-291 (Bird -> Dog) (test)
11.29FID
6
ImageNet (val)
71.75CH
6
ImageNet-1K (test)
100ASR (IncRes-v2)
6
ImageNet (test)
0LPIPS
6
ImageNet-1K
0.889Accuracy
6