zjhhhh/stage3_mean_beta_0.1_multi_expand_tokenized

这是一个包含多个配置文件的数据集，每个配置文件定义了数据集的训练集和测试集路径。数据集的每个实例包含多个字段，包括字符串类型的提示、要求、各种响应，以及浮点数列表类型的各种均值和多数值。此外，还包括整数类型的字段和字符串类型的提示及其分词结果。数据集的规模较大，包含超过2800万条训练数据和700多条测试数据。

This is a dataset consisting of multiple configuration files, where each configuration file defines the paths of the training and test datasets. Each instance in the dataset comprises multiple fields, including string-type prompts, requirements, various responses, as well as various mean and majority values in the form of float lists. Additionally, it includes integer-type fields, as well as string-type prompts and their tokenization results. The dataset has a large scale, containing over 28 million training instances and more than 700 test instances.