NSFW Content Moderation on Malicious NSFW datasets benchmark leaderboard · SOTA2 Research

NSFW Content Moderation on Malicious NSFW datasets

1.5Unsafe Ratio (Sexually Explicit)PromptGuard

Evaluation Results

Method	Links
PromptGuard2025.01		1.5	5.17	12.17	4.5	5.84
SafeGen2025.01		2.2	24.5	38.67	12.83	19.1
UCE2025.01		4.33	8.17	29.83	7.83	12.54
SafetyFilter2025.01		15.67	25.33	32.17	16.17	22.34
SLD Max2025.01		36.33	9.67	37.33	8.33	22.92
SLD Strong2025.01		41.83	13.83	35.67	8.33	24.92
POSI2025.01		45.17	18.5	34.67	13.17	27.88
SDv2.12025.01		45.67	33.83	38.83	19.67	34.5
SDv1.42025.01		71.17	30	36.17	19.5	39.21