math-extraction-comp/Goekdeniz-Guelmez__Josiefied-Qwen2.5-7B-Instruct-abliterated-v2

该数据集包含多个字段，其中包括问题（question）、参考答案（gold）、目标答案（target）、预测答案（prediction）等字符串类型的字段，以及多个评分相关字段，如lighteval-4cfbbf17_score等浮点数类型的字段。数据集被拆分为训练集（train），包含1324个示例，总大小为3,811,618字节。提供了默认配置，指定了训练集的数据文件路径。

The dataset contains multiple fields, including string-type fields such as question, reference answer (gold), target answer (target), predicted answer (prediction), and several scoring-related fields like lighteval-4cfbbf17_score which are of float64 type. The dataset is split into a training set (train) with 1,324 examples, totaling 3,811,618 bytes in size. A default configuration is provided, specifying the data file path for the training set.