1 paper
Tianyi Huang, Nathan Huang, Justin Tang +2
Large language models (LLMs) are now widely used as judges, yet their decisions can change under presentation choices that should be irrelevant. We study one such source of instabi…