Estimating the Self-Consistency of LLMs

Robert Nowak

Systems often repeat the same prompt to large language models (LLMs) and aggregate responses to improve reliability. This short note analyzes an estimator of the self-consistency o…