hooriehsabzevari/Daniar_tokenized__CLUSTER

该数据集包含训练和测试两个分割，分别有8390和933个样本。每个样本包含多个特征，如chosen和reject，它们都是列表类型，包含content（字符串类型）和role（字符串类型）两个子特征。此外，还包括chosen_token和reject_token（int64序列）、chosen_mask和reject_mask（int64序列）、chosen_reward和reject_reward（float64类型）、以及chosen_logprob和reject_logprob（float64类型）。这些特征可能用于比较或评估模型在不同选择上的表现。

The dataset includes train and test splits with 8,390 and 933 examples, respectively. Each example contains multiple features such as chosen and reject, which are lists with sub-features content (string type) and role (string type). Additionally, it includes chosen_token and reject_token (int64 sequences), chosen_mask and reject_mask (int64 sequences), chosen_reward and reject_reward (float64 type), and chosen_logprob and reject_logprob (float64 type). These features are likely used for comparing or evaluating model performance on different choices.