AI.info

Dongxia Wang

Explore Dongxia Wang on AI.info.

• LLM-VA: Resolving the Jailbreak-Overrefusal Trade-off via Vector Alignment