ImageNet
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ImageNet (val)
73.8Top-1 Accuracy
4
ImageNet (val)
58.8Model Size (Mbits)
4
ImageNet (val)
47.3Loc Acc
4
ImageNet 3-hop Unseen & seen categories
2.9Top-1 Accuracy
4
ImageNet 2-hop (Unseen & seen categories)
10.5Top-1 Accuracy
4
ImageNet 32x32 downsampled (val)
3.77NLL (bits/dim)
4
ImageNet 128x128 (test)
42.07FID
4
ImageNet unseen classes (test)
8.7Top-1 Accuracy (Per-Class)
4
ImageNet (val)
3.77NLL (bits/dim)
4
ImageNet L_dogs full (train)
99Accuracy
4
ImageNet L_dogs full (test)
99Accuracy
4
ImageNet #L_rand full (train)
99Accuracy
4
ImageNet L_rand full (test)
99Accuracy
4
ImageNet 32x32 (train)
3.77NLL (bits/dim)
4
ImageNet All
1.5Flat Hit@1
4
ImageNet 3-hop
2.9Flat Hit@1
4
ImageNet 2-hop
10.5Flat Hit@1
4
miniImageNet (test)
2Convergence Time (67% Acc, hours)
4
ImageNet 64 x 64 (train)
2.16FID
4
ImageNet 256p 1k (val)
3.02gFID
4
Tiny-ImageNet
2.62ESS/s
3
ImageNet-1K (val)
5.005Clean FPR
3
ImageNet-1K (val)
84.2Top-1 Accuracy (Original)
3
ImageNet 256x256 (σ=0.3) (val)
29.13PSNR
3
ImageNet 256x256 (σ=0.2) (val)
30.66PSNR
3