1 paper
Xiao Li, Joel Kreuzwieser, Alan Peters
We investigate how large language models respond to prompts that differ only in their token-level realization but preserve the same semantic intent, a phenomenon we call prompt var…