LuckyLukke/wordle_REFUEL_turnwise_

这是一个包含游戏、实验和剧集信息的NLP数据集。数据集包含选择的文本内容、是否有错误、角色信息、拒绝的文本内容、拒绝的错误信息、拒绝的角色信息、选择和拒绝的奖励值、选择和拒绝的词标记序列、选择和拒绝的掩码序列以及选择和拒绝的对数概率。数据集分为训练集和测试集，分别包含2596和300个示例。

This is an NLP dataset containing information about games, experiments, and episodes. The dataset includes chosen text content, error presence, role information, rejected text content, rejection error information, rejection role information, reward values for chosen and rejected, token sequences for chosen and rejected, mask sequences for chosen and rejected, and log probabilities for chosen and rejected. The dataset is split into a training set and a test set, containing 2596 and 300 examples respectively.