1 paper · 1 filter
Warren Li, Yiqian Wang, Zihan Wang +1
In-context learning (ICL) enables large language models to perform new tasks by conditioning on a sequence of examples. Most prior work reasonably and intuitively assumes that whic…