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

P4D

Benchmarks

Task NameDataset NameSOTA ResultTrend
NSFW suppressionP4D
ASR93.4
41
Concept ErasureP4D
ASR98.7
23
Defense against adversarial promptsP4D (test)
Attack Success Rate1.1
15
Concept UnlearningP4D
P4D0.0027
14
Safe Text-to-Image GenerationP4D
ASR98.7
13
Nudity ErasureP4D
Unsafe Rate16.17
7
Image GenerationP4D
Nudity Rate2
2
Text-to-Image Synthesis SafetyP4D (Prompt4Debugging)
Total Violation Count16
2
Showing 8 of 8 rows