Datasets used to train the LION pipeline. Paper: https://arxiv.org/abs/2407.06542; Code: https://github.com/Columbia-NLP-Lab/LionAlignment