An Empirical Analysis of Uncertainty in Large Language Model Evaluations

Qiujie Xie, Qingqiu Li, Zhuohao Yu +3

As LLM-as-a-Judge emerges as a new paradigm for assessing large language models (LLMs), concerns have been raised regarding the alignment, bias, and stability of LLM evaluators. Wh…