yguooo/summarize_from_feedback_tldr_3_filtered_oai_preprocessing_llama3_scene3

该数据集是用于OpenAI的Summarize from Feedback任务的TL;DR SFT数据集，包含Reddit帖子的标题、内容、摘要等信息。数据集经过预处理，生成了用于训练的查询和参考响应，并分为训练集、验证集和测试集。数据集的特征列包括帖子的唯一标识、子版块、标题、正文、摘要、参考响应等，并添加了预处理生成的查询、查询的token化版本、参考响应的token化版本等列。

This dataset is directly taken from OpenAIs Summarize from Feedback task, primarily used for generating summaries of Reddit posts. The dataset includes multiple fields such as id, subreddit, title, post, summary, etc., which are directly sourced from the original dataset. Additionally, some fields like query, query_token, etc., are added by the preprocessing script, which are used for query and response generation in the summarization task. The dataset is divided into training, validation, and test sets, each with corresponding sample counts and byte sizes.