imminsik/summarize_from_feedback_tldr_3_filtered_oai_preprocessing_gpt2_1757477914

该数据集是从Reddit平台获取的，包含了帖子的唯一标识符、所属子版块、标题、帖子内容、摘要以及参考回复。此外，预处理脚本添加了用于摘要的长度限制查询、查询的标记化版本、参考回复的标记化版本、参考回复标记化长度、查询和参考回复的连接以及该连接的标记化版本和长度。数据集分为训练集、验证集和测试集，用于OpenAI的Summarize from Feedback任务。This dataset is from Reddit and includes unique identifier for the post, the subreddit it was taken from, the title of the post, the body of the post, the summary, and the reference response. Additionally, the preprocessing script adds a length-limited query for summarization, the tokenized version of the query, the tokenized version of the reference response, the length of the tokenized reference response, the concatenation of the query and the reference response, and the tokenized version of this concatenation and its length. The dataset is split into training, validation, and test sets for OpenAIs Summarize from Feedback task.