For classification settings, you may want to try accuracy, precision, recall, F1 score, and AUC ROC. For explainable-specific metrics, you can try methods like LIME and SHAP. Hope this is helpful.