Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

VLGuard & MSSBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Malicious prompt detectionVLGuard & MSSBench π = 0.005
AUROC (VLGuard & MSSBench)0.9195
26
Malicious Prompt DetectionVLGuard & MSSBench
AUROC97.48
5
Malicious Content DetectionVLGuard & MSSBench
Typographic AUROC95.65
3
SafeguardingVLGuard & MSSBench (test)
Average Score88.71
3
Showing 4 of 4 rows