Distillability of LLM Security Logic: Predicting Attack Success Rate of Outline Filling Attack via Ranking Regression

Open in new window