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

Malicious Prompt Detection on VLGuard & MLLMGuard π = 0.005

94.37AUROC

VLMGuard

49.587661.213872.8484.4662Oct 1, 2024
Updated 18d ago

Evaluation Results

MethodLinks
2024.10
94.3794.79
2024.10
91.9493
2024.10
88.8788.53
2024.10
86.4189.4
2024.10
82.1183.86
2024.10
81.2785.33
2024.10
80.5282.7
2024.10
78.2982.51
2024.10
76.8778.32
2024.10
76.1277.1
2024.10
73.7871.4
2024.10
73.5780.56
2024.10
71.7778.55
2024.10
70.3476.57
2024.10
65.0772.52
2024.10
63.7771.14
2024.10
62.7559.62
2024.10
61.4169.57
2024.10
61.0861.34
2024.10
61.0559.99
2024.10
60.5162.15
2024.10
59.2759.44
2024.10
54.1963.68
2024.10
52.6857.25
2024.10
52.3161.4
2024.10
51.3154.8