Rapidata/text-2-video-human-preferences-kling-v2.1-master

该数据集包含来自约20,000名人类注释者的约60,000个响应，用于评估Kling v2.1 Master视频生成模型。数据集包括提示、视频链接、对齐、一致性和偏好方面的加权结果、包含用户信息的详细结果、文件名和模型名称等特征。它还指定了数据集的大小、语言和类别。该数据集是使用Rapidata Python API收集的，旨在用于大规模数据注释。README进一步解释了列的内容，并提供了对齐、一致性和偏好分数的示例。最后，它提供了有关Rapidata的信息，并链接到其他数据集。

This dataset includes approximately 60,000 responses from around 20,000 human annotators to evaluate the Kling v2.1 Master video generation model. The dataset features prompts, video links, weighted results for alignment, coherence, and preference, detailed results with user information, and filenames and model names. It also specifies the datasets size, language, and categories. The dataset was collected using the Rapidata Python API and is designed for large-scale data annotation. The README further explains the columns, providing examples for alignment, coherence, and preference scores. Lastly, it provides information about Rapidata and links to other datasets.