Zhixuan Chu

The paper proposes a black-box attack using a hierarchical genetic algorithm to induce 'overthinking' in Large Reasoning Models, demonstrating that this vulnerability can cause significant resource exhaustion.

The paper introduces StreamSynth, a sequential setting for synthetic data generation, and proposes SynLearner, a framework that enables LLMs to improve synthesis performance by accumulating and transferring experience across a stream of tasks.

The paper introduces ConsisGuard, a framework that addresses the 'deliberation-to-enforcement gap' in LLM guardrails by ensuring that the reasoning process is faithfully and consistently translated into the final safety decision.