ZixuanKe/flare_finqa_sup_sample_from_policy_v1.1_stepwise_dpo_binarized

该数据集包含多个特征字段，包括prompt（提示）、rejected（被拒绝的响应）、chosen（被选中的响应）、justification（理由）以及几个与llama3模型相关的长度字段。数据集分为训练集和验证集，训练集包含32,843个样本，验证集包含1,724个样本。总下载大小为36,344,572字节，数据集总大小为186,839,938字节。

The dataset includes multiple feature fields such as prompt, rejected, chosen, justification, and several length fields related to the llama3 model. The dataset is divided into a training set with 32,843 examples and a validation set with 1,724 examples. The total download size is 36,344,572 bytes, and the total dataset size is 186,839,938 bytes.