MAGNET Dataset: Text Samples Used in the Study "Counterfactual Samples Synthesizing for Mitigating Hallucination in Large Language Models"

This dataset contains the text samples used in the study "MAGNET: Counterfactual samples synthesizing for mitigating hallucination in large language models."The dataset includes input–output text pairs that were used to evaluate and analyze the proposed counterfactual sample synthesis method (MAGNET).No additional model parameters, code, or evaluation results are included.The data are provided to support transparency and reproducibility of the study.