ImageNet
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ImageNet 1k (val)
83.9Top-1 Acc (W6A6)
23
Tiny-ImageNet (test)
99.65Residual Accuracy
23
ImageNet
1.39Expected Calibration Error
23
ImageNet (val)
1,840ViT Performance
23
ImageNet 256x256 1.0 (train val)
1.42FID
23
ImageNet In-Distribution (test)
86.35ID Accuracy
23
ImageNet OOD variants (A, V2, R, S) 1k (test val)
77.87Top-1 Acc (ImageNet)
23
ImageNet
82.63Base Accuracy
23
ImageNet 1K (val)
86.2FT Accuracy
23
Tiny-ImageNet (test)
0.016ECE
23
ImageNet-R
-5.61IN-R Gap
23
ImageNet-R (test)
89.9Accuracy
23
ImageNet-LT
70.09Top-1 Acc (Forward-LT, IR=50)
23
ImageNet 1k (val)
86Accuracy
23
MiniImageNet 33.3% symmetric label swap noise 54 (test)
56.89Accuracy
23
MiniImageNet 0% noise 54 (test)
64.28Accuracy
23
ImageNet-100 10 steps
95.1Top-5 Accuracy (Last)
23
ImageNet-A Target
51.8Accuracy
23
ImageNet (val)
79Accuracy
23
ImageNet 1k (val)
75.5Top-1 Accuracy
23
ImageNet 10% label fraction
95.5Top-5 Acc
23
ImageNet-32
4.11FID
23
ImageNet-100 50% Compression (test)
87.18Top-1 Accuracy
22
ImageNet
77.43Accuracy (Base)
22
ImageNet (val)
5.21Deletion
22