While Large Language Models (LLMs) have proven to be exceptional on a variety of tasks after alignment, they may still produce responses that contradict the context or world knowledge confidently, a phenomenon known as “hallucination”.



Source link