LM Contamination Index

该数据集定期更新，对一系列开放和专有模型的污染程度进行估算。此外，该指数通过零样本提示模型生成特定数据集的实例，详细说明了所需的分割和格式，以便评估语言模型中的数据污染情况。

This dataset is periodically updated to estimate data contamination levels across a range of open and proprietary models. Furthermore, this index generates instances of specific datasets by zero-shot prompting models, and specifies the required data splits and formats to facilitate the evaluation of data contamination in language models.