CLIP (zero shot)

Reported on 3 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology3 results

Anomaly DetectiononLeave-One-Class-Out CIFAR-10
AUROC· 2022-05-23
92.2
best: 98.4 (BCE-CLIP)
Exposing Outlier Exposure: What Can Be Learned From Few, One, and Zero Outlier Images arXiv:2205.11474
Anomaly DetectiononLeave-One-Class-Out ImageNet-30
AUROC· 2022-05-23
97.8
best: 99.3 (BCE-CLIP (OE))
Exposing Outlier Exposure: What Can Be Learned From Few, One, and Zero Outlier Images arXiv:2205.11474
Anomaly DetectiononOne-class CIFAR-10
AUROC· 2022-05-23
98.5
best: 99.6 (CLIP (OE))
Exposing Outlier Exposure: What Can Be Learned From Few, One, and Zero Outlier Images arXiv:2205.11474