dsrtrain/new_script_raft_iter4_gen_with_rewards
收藏官方服务：
资源简介：
这是一个包含索引、提示、答案序列、地面真实和风格信息的结构化数据集，以及一个表示奖励的布尔序列。数据集被分割为训练集，其中包含20000个示例，文件大小为280259157字节。
This is a structured dataset containing index, prompt, answer sequence, ground truth and style information, as well as a boolean sequence representing rewards. The dataset is split into a training set, which includes 20000 examples and has a file size of 280259157 bytes.
提供机构：
dsrtrain
