yuxi-liu-wired/stylebreeder_2M_embedding

这是一个基于stylebreeder/stylebreeder数据集的版本，包含了2M的子集。每个图像都经过CSD模型处理，生成了两个768维的嵌入向量（风格嵌入和内容嵌入），并且图像以JPEG格式保存以节省空间。数据集包含三个主要字段：style_embedding、content_embedding和image。由于上传时的问题，数据集查看器无法正常工作。

This is a version of [stylebreeder/stylebreeder](https://huggingface.co/datasets/stylebreeder/stylebreeder), the 2M split. The dataset includes two parts: style_embedding and content_embedding, each being a 768-dimensional sequence of float16. Additionally, the dataset contains image data, which are stored in JPEG format with a much smaller resolution to save space. The creation process of the dataset involves running the CSD model on each image, resulting in two embedding vectors.