The prompt given by users to the LLM to generate inferences.