Zhuoyan Xu

A new method enhances attention mechanisms in language models for better performance.

Examining why larger models struggle with in-context learning compared to smaller ones.

Exploring how LLMs perform on composite tasks that combine simpler tasks.