Ruihan Wu

Learning-Time Encoding Shapes Unlearning in LLMs

Private-RAG: Answering Multiple Queries with LLMs while Keeping Your Data Private

Can We Infer Confidential Properties of Training Data from LLMs?

On Speeding Up Language Model Evaluation

Do LLMs Really Forget? Evaluating Unlearning with Knowledge Correlation and Confidence Awareness

Large Scale Knowledge Washing

Evaluating Deep Unlearning in Large Language Models

Influence-based Attributions can be Manipulated