behavior-in-the-wild/UltraLAMBDA

UltraLAMBDA是一个大规模数据集，包含从YouTube和Facebook Ads等平台以及CommonCrawl获取的广告数据。这些广告的记忆性评分由名为Henry的模型给出。数据集的每个条目包含视频的YouTube ID和记忆性标签（低、中、高）。UltraLAMBDA is a large-scale dataset of ads sourced from brand videos on platforms such as YouTube and Facebook Ads, as well as from CommonCrawl. The memorability scores for the ads are assigned by our model Henry. The dataset includes two main fields: id, representing the YouTube ID for the video; memorability, representing the memorability label for the video (low, medium, or high). The dataset is divided into a training set, containing 190041 samples, with a total size of 4424635 bytes.