prli/pubmed-full_chopped_pyt_llama_alpha_0-1

该数据集包含文本、元信息、句子扰动数量以及文档统计信息。文本是数据集的主要特征，元信息中包含了数据集的集合名称。文档统计信息详细记录了文档的优势分数、长度变化比例、原始句子长度、替换后的句子长度等多个统计数据。数据集分为验证集，验证集大小为4000个样本。

The dataset includes text, metadata, the number of perturbed sentences, and document statistics. Text is the primary feature, and metadata contains the set name of the dataset. Document statistics detail various statistics such as advantage scores, length change ratio, original sentence length, and replacement paraphrase length. The dataset is split into a validation set, which contains 4000 samples.