yguooo/summarize_from_feedback_oai_preprocessing_llama3_scene1

该数据集包含多个特征，如info、summaries、choice、worker等，每个特征下又有多个子字段。数据集分为train、validation和validation_cnndm三个部分，每个部分都有对应的文件大小和示例数量。具体来说，info特征包含id、post、title、subreddit、site和article等字段；summaries特征包含text、policy和note等字段；choice特征为整型数据；worker和batch特征为字符串类型；split特征表示数据集的分割类型；extra特征包含confidence字段；query_token、chosen_token、rejected_token等特征为序列类型；query、chosen、rejected等特征为字符串类型；policies特征为字符串类型；query_chosen、query_rejected等特征为字符串类型；query_token_len、query_chosen_token_len、query_rejected_token_len等特征为整型数据；query_chosen_token_response_label和query_rejected_token_response_label为序列类型。数据集的总下载大小为294917146字节，总数据集大小为6232212662字节。

The dataset contains multiple features such as info, summaries, choice, worker, etc., each with multiple subfields. The dataset is divided into three parts: train, validation, and validation_cnndm, each with corresponding file sizes and example counts. Specifically, the info feature includes fields such as id, post, title, subreddit, site, and article; the summaries feature includes fields such as text, policy, and note; the choice feature is of integer type; the worker and batch features are of string type; the split feature indicates the type of dataset split; the extra feature includes the confidence field; the query_token, chosen_token, rejected_token features are of sequence type; the query, chosen, rejected features are of string type; the policies feature is of string type; the query_chosen, query_rejected features are of string type; the query_token_len, query_chosen_token_len, query_rejected_token_len features are of integer type; the query_chosen_token_response_label and query_rejected_token_response_label features are of sequence type. The total download size of the dataset is 294917146 bytes, and the total dataset size is 6232212662 bytes.