CohenQu/POPE-93-medium-first_guide-no_guide

该数据集包含多个特征，包括数据来源（data_source）、提示（prompt，包含内容和角色）、能力（ability）、奖励模型（reward_model，包含真实值和风格）、额外信息（extra_info，包含索引和分割）、等级（level）和问题ID（problem_id）。数据集分为训练集（256个样本）和测试集（180个样本），总大小约为355KB。The dataset includes multiple features such as data source (data_source), prompt (containing content and role), ability, reward model (containing ground truth and style), extra information (containing index and split), level, and problem ID (problem_id). The dataset is divided into a training set (256 examples) and a test set (180 examples), with a total size of approximately 355KB.