Jun Sun

A new method for assessing language models' alignment with human values.

Improving machine learning robustness against adversarial examples is critical for secure applications.

This framework uses multiple agents and task graphs for efficient problem solving.

A new method to edit language models effectively while maintaining performance.

This study investigates the limits of accuracy and robustness in machine learning models.

TlSbTe shows promise in thermoelectric applications with low thermal conductivity and favorable electrical properties.

CCL ensures neural networks maintain accuracy while learning new tasks.

Learn how surfaces change while keeping their area constant.

A new method helps protect language models from harmful backdoor attacks.

Bridging Large Language Models and Formal Methods for reliable AI solutions.

Explore the fascinating world of space curvature and its implications.