DOGe: Defensive Output Generation for LLM Protection Against Knowledge Distillation

Open in new window