Weilun Xu Alun

I approach this question through machine learning, mechanistic interpretability and cognitive science. I study the causal role of internal representations in reasoning and behaviour by developing computational models and intervention methods. I also want to understand how architectural choices influence what a system learns and how it reasons.