List your company

Safety Control on BlackSheep Llama3.2-3B

93.5Safety-Quality Score (P_safeguarded)raw_history
15.70835.90456.176.296May 25, 2026

Evaluation results

MethodLinks
raw_historyModel Type=Hand-crafted2026.0593.50.742
hybridModel Type=Hand-crafted2026.0593.50.742
best_worst_recentModel Type=Hand-crafted2026.0590.80.715
performance_tieredModel Type=Hand-crafted2026.0587.30.68
ai_onlyModel Type=Hand-crafted2026.0587.20.68
progressiveModel Type=Hand-crafted2026.0587.20.68
ai_enhancedModel Type=Hand-crafted2026.0586.90.676
minimalModel Type=Hand-crafted2026.0586.70.674
contrast_learningModel Type=Hand-crafted2026.0584.60.654
SafeCtrl-RLModel Type=Our Proposed2026.0583.30.641
smart_adaptiveModel Type=Hand-crafted2026.0577.50.582
trajectory_learningModel Type=Hand-crafted2026.05750.557
OPROModel Type=Prompt Optimisation2026.0555.30.36
DynamicRetrievalModel Type=Prompt Optimisation2026.0539.30.2
GRIPSModel Type=Prompt Optimisation2026.0538.70.194
TextGradientModel Type=Prompt Optimisation2026.0525.80.065
EvolutionaryModel Type=Prompt Optimisation2026.0518.7-0.006