1 paper · 1 filter
Chenrui Liu, Falong Tan, Chuanlong Xie +2
This paper investigates the expected excess risk of in-context learning (ICL) for multiclass classification. We formalize each task as a sequence of labeled examples followed by a…