List your company

Sponge Attack Detection on RL-GOAL

99.6F1 ScoreSHIELD
93.474495.064796.65598.2453Jan 27, 2026

Evaluation results

MethodLinks
SHIELDTarget Model=LLaMA22026.0199.6
Perplexity-filterTarget Model=LLaMA22026.0199.19
Harm-filterTarget Model=LLaMA22026.0193.71