wittenator/celeba-hq-256x256
收藏资源简介:
该数据集名为madebyollin~sdxl-vae-fp16-fix,是一个用于机器学习的图像数据集,专门针对SDXL(Stable Diffusion XL)变分自编码器(VAE)的fp16修复版本。数据集包含两个主要特征:encoded_image,这是一个形状为[4, 32, 32]的浮点型三维数组,表示经过编码的图像数据;label,这是一个类标签,包含两个类别:0和1,可能用于二元分类任务。数据集分为训练集和验证集,训练集有28000个样本,验证集有2000个样本,总下载大小约为510.84 MB,数据集大小约为507.72 MB。数据文件路径配置在configs部分指定,适用于模型训练和验证。
The dataset named madebyollin~sdxl-vae-fp16-fix is a machine learning image dataset specifically designed for the fp16 fix version of the SDXL (Stable Diffusion XL) Variational Autoencoder (VAE). It includes two main features: encoded_image, which is a float32 three-dimensional array with a shape of [4, 32, 32], representing encoded image data; and label, a class label with two categories: 0 and 1, likely used for binary classification tasks. The dataset is split into a training set with 28,000 examples and a validation set with 2,000 examples. The total download size is approximately 510.84 MB, and the dataset size is approximately 507.72 MB. Data file paths are configured in the configs section, suitable for model training and validation.




