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

Safety Evaluation

Benchmarks

Task NameDataset NameSOTA ResultTrend
Safety AlignmentMultiple-choice safety evaluation
Refusal Rate84.48
11
Safety EvaluationSafety Evaluation
Overall Safety Score73.97
9
Safety DefenseSafety Evaluation Text-only
Training Time (hr)0.4
5
Safety DefenseSafety Evaluation Multimodal
Inference Overhead (Multiplier)1
4
Safety EvaluationSafety Evaluation
Illegal Content Count0
3
Safety DetectionSafety Evaluation Strict Safety Mode
Precision50
1
Showing 6 of 6 rows