List your company

Attack Detection on Mixed Attacks 105K sample set

77Detection RateLogReg (Ours)
37.27247.58657.968.214Feb 15, 2026

Evaluation results

MethodLinks
LogReg (Ours)input=raw activations, evaluation_regime=LODO2026.0277
Llama-as-Judgeacronym=LJ, prompting=zero-shot2026.0273.8
PromptGuard 2acronym=PG2026.0254.4
LlamaGuardacronym=LG2026.0238.8