caedencode/caeden-instruct-2.0-merged-dataset

该数据集包含多个特征字段，如id、标题、上下文、问题、答案等，其中答案字段包含文本和答案起始位置两个子字段。数据集分为一个训练集，包含638,414个样本，总大小为461,364,435字节。下载大小为203,955,159字节。数据集的配置文件中指定了默认配置，数据文件路径为data/train-*。

The dataset includes multiple features such as id, title, context, question, answers, system_prompt, output, instruction, text, and input. The answers feature is a sequence containing text and answer_start sub-features. The dataset is divided into a training set with 638414 samples. The download size of the dataset is 203955159 bytes, and the dataset size is 461364435 bytes.