Tao Liang

SPEED: Scalable, Precise, and Efficient Concept Erasure for Diffusion Models

SAFER: Probing Safety in Reward Models with Sparse Autoencoder

Enhancing Multi-Modal LLMs Reasoning via Difficulty-Aware Group Normalization

SPEED: Scalable, Precise, and Efficient Concept Erasure for Diffusion Models

Grounding Language with Vision: A Conditional Mutual Information Calibrated Decoding Strategy for Reducing Hallucinations in LVLMs

Neuron-Level Sequential Editing for Large Language Models