Drift No More? Context Equilibria in Multi-Turn LLM Interactions

Vardhan Dongre, Ryan A. Rossi, Viet Dac Lai +3

Large Language Models (LLMs) excel at single-turn tasks such as instruction following and summarization, yet real-world deployments require sustained multi-turn interactions where…