grayapple/SIKO

该数据集是为研究目的创建的，旨在反映韩语在语言模型中的句法特征。它基于KLUE-benchmark数据集，并遵循其许可条款。数据集通过改变词序、删除格标记等方式生成，具体包括语义保留的词序改变、非语义保留的词序改变、格标记删除以及混合数据。数据示例展示了SIKO-NLI和SIKO-TC的具体数据格式和内容。

This dataset was created for research purposes to reflect the syntactic characteristics of the Korean language in language models. It is based on the KLUE-benchmark dataset and adheres to the licensing terms of the KLUE-benchmark. The dataset includes various processing methods such as semantic-preserving and non-semantic-preserving word order changes, case marker deletions, and mixed data. The dataset contains 24.9k data for Text Classification and 30k data for NLI. Data examples are provided for SIKO-NLI and SIKO-TC.