Faithfulness Hallucination

The phenomenon in which a large model's answer appears plausible but deviates from the user's instructions or the input context. Its root cause lies in insufficient instruction-following capability of the model, which requires improvement through fine-tuning or RLHF.