Fengqing Jiang

large language model (8) jailbreak attack (3) safety alignment (3) adversarial attack (2) instruction tuning (2) knowledge distillation (2) chain-of-thought reasoning (2) decoding strategy (2) llm safety (2) adversarial learning (2) model safety (1) safety evaluation (1) adversarial training (1) adversarial prompt (1) reasoning trace (1) backdoor attack (1) reasoning benchmark (1) token probability (1) reasoning model (1) harmful content (1)