List your company

Multi-modal Toxicity Attack on RealToxicityPrompts (RTP) (test)

31.36Overall ScoreSTARE (0.2)
4.153611.216818.2825.3432May 1, 2026

Evaluation results

MethodLinks
STARE (0.2)Access Tier (Tier-1 vs Tier-2)=Tier-1, Alignment weight=0.22026.0531.3617.16.2929.734.3815.956.140.78
STARE (0.05)Access Tier (Tier-1 vs Tier-2)=Tier-1, Alignment weight=0.052026.0528.9316.845.7226.573.916.695.770.72
STARE (0.1)Access Tier (Tier-1 vs Tier-2)=Tier-1, Alignment weight=0.12026.0528.5417.085.7725.754.2216.146.820.76
STARE w/ DDPOAccess Tier (Tier-1 vs Tier-2)=Tier-1, Access assumption (cross-paradigm vs controlled)=controlled; identical white-box T2I budget as STARE, Configuration=with DDPO2026.0527.8415.625.4126.123.9215.115.80.75
STARE w/o AlignAccess Tier (Tier-1 vs Tier-2)=Tier-1, Component removal (Ablation)=w/o Align2026.0526.4314.845.0925.83.9514.055.830.68
STARE w/o EditAccess Tier (Tier-1 vs Tier-2)=Tier-1, Component removal (Ablation)=w/o Edit2026.0525.5613.464.6824.883.2710.565.50.72
STARE w/o LoRAAccess Tier (Tier-1 vs Tier-2)=Tier-1, Component removal (Ablation)=w/o LoRA2026.0522.0412.135.1119.774.0911.065.990.75
ARTAccess Tier (Tier-1 vs Tier-2)=Tier-2, Access assumption (cross-paradigm vs controlled)=cross-paradigm; different access, for context2026.0518.629.222.1517.544.058.946.450.75
DiffZOOAccess Tier (Tier-1 vs Tier-2)=Tier-2, Access assumption (cross-paradigm vs controlled)=cross-paradigm; different access, for context2026.0517.29.012.1216.422.727.884.140.73
PGJAccess Tier (Tier-1 vs Tier-2)=Tier-2, Access assumption (cross-paradigm vs controlled)=cross-paradigm; different access, for context2026.0514.867.851.7513.981.658.093.430.71
Text + SDAccess Tier (Tier-1 vs Tier-2)=Tier-2, Access assumption (cross-paradigm vs controlled)=cross-paradigm; different access, for context2026.0511.155.710.3810.630.226.113.970.72
Text-OnlyAccess Tier (Tier-1 vs Tier-2)=Tier-2, Access assumption (cross-paradigm vs controlled)=cross-paradigm; different access, for context2026.055.23.10.25.10.82.80.6—