LongVILA/longvila_sft_dataset

该数据集包含视频对话数据，每个数据项包括一个唯一的id、一个视频链接和一个对话列表。对话列表中的每个对话包含发言者（from）和发言内容（value）。数据集仅包含训练集，共有70985个样本，总大小为54940081字节，下载大小为28482909字节。

This dataset contains video conversation data, with each data item including a unique id, a video link, and a list of conversations. Each conversation in the list includes a speaker (from) and the content of the speech (value). The dataset only includes a training set, with a total of 70,985 samples, a total size of 54,940,081 bytes, and a download size of 28,482,909 bytes.