Modeling and Predicting Multi-Turn Answer Instability in Large Language Models

Jiahang He, Rishi Ramachandran, Neel Ramachandran +5

As large language models (LLMs) are adopted in an increasingly wide range of applications, user-model interactions have grown in both frequency and scale. Consequently, research ha…