Rapidata/text-2-video-human-preferences-genmo-mochi-1

这个数据集包含了从约20k人类注释者那里收集的约60k人类响应，用于评估mochi-1视频生成模型。数据集使用Rapidata Python API在约30分钟内收集，任何人都可以访问，非常适合大规模数据标注。数据集包括成对的视频比较，每个条目包含video1和video2字段，这些字段包含链接到缩放后的GIF图像，以便于查看。完整分辨率的视频可以在这里找到。加权结果列包含从0到1的分数，代表聚合的用户响应。单个用户响应可以在详细结果列中找到。数据集还包含对齐、一致性和偏好的评分。

This dataset contains approximately 60k human responses collected from about 20k human annotators to evaluate the mochi-1 video generation model. The dataset was collected in about 30 minutes using the Rapidata Python API, which is accessible to anyone and ideal for large-scale data annotation. The dataset includes paired video comparisons, with each entry containing video1 and video2 fields, which include links to downsampled GIF images for easy viewing. The full-resolution videos can be found here. The weighted_results column contains scores ranging from 0 to 1, representing aggregated user responses. Individual user responses can be found in the detailedResults column. The dataset also includes alignment, coherence, and preference scores.