Training Data Extraction Challenge

该数据集是从网络来源搜集的，用于评估摘要片段的可提取性。此外，它还用于针对模型复述行为的定向评估，目的是检测模型在提取性表现方面的能力。

This dataset is collected from web sources to evaluate the extractability of summary segments. Furthermore, it is also used for targeted evaluations of model paraphrasing behaviors, aiming to detect models' capabilities in extractive performance.