Ying Liu

The paper proposes SRTJ, a Self-Evolving Rule-Driven Training-Free Jailbreak framework that systematically discovers and refines attack strategies using rule composition and feedback to achieve robust and generalizable jailbreaking against modern LLMs.

The paper introduces MTAVG-Bench 2.0, a new benchmark designed to diagnose high-level failure modes of cinematic expressiveness in multi-talker audio-video generation, showing that even advanced models struggle with complex scene-level failures.

The paper proposes a novel framework combining behavior-invariant task representation learning and a Transformer-based world model to achieve robust generalization in offline meta-reinforcement learning, particularly in sparse-reward settings.

The paper identifies a failure mode called spatial lexical bias in MLLMs, where adding a spatial word to options biases the model's choice, and demonstrates that this failure originates primarily from the language processing side rather than poor visual attention.