small models cannot encode so many facts, they will hallucinate more out-of-box a key method to help with hallucinations is to provide good sources when asking

small models cannot encode so many facts, they will hallucinate more out-of-box a key method to help with hallucinations is to provide good sources when asking questions (context engineering / knowledge base)