ashmitkx/content-behavior-corpus

该数据集包含视频ID、自动语音识别（ASR）原始数据、ASR分组数据、光学字符识别（OCR）数据、BLIP2注释、重播图、元数据和频道数据等特征。数据集分为训练集和测试集，训练集包含22569个样本，测试集包含2026个样本。The dataset includes multiple features such as video ID, ASR raw data, ASR grouped, OCR, BLIP2 annotations, replay graphs, metadata, and channel data. Each feature has detailed sub-features and data types. The dataset is divided into train and test splits, each containing different numbers of bytes and examples.