Imported from https://github.com/MI2DataLab/nlp_interpretability_framework