Datasets from "A Novel Metric for Measuring the Robustness of Large Language Models in Non-adversarial Scenarios" (https://arxiv.org/abs/2408.01963)