A prompting method that reduces hallucinations by grounding responses in pre-training data. Update the source and question variable to run this prompt.
