CVAE-CLIP Project

Overview

This project contains code for training and evaluating a Conditional Variational Autoencoder (CVAE) model on the CelebA dataset. The primary goal is to generate images conditioned on specific attributes using the CVAE model.

Environment Setup and Dependencies

To set up the environment and install the required dependencies, follow these steps: Clone the repository:

git clone https://github.com/laowangshini/CVAE-CLIP.git
cd CVAE-CLIP

Repository Structure

The repository contains the following files and directories:

• realtest2_clipcvae.ipynb: Jupyter notebook for training and evaluating the CVAE model on the CelebA dataset.
• realtest2_cvae.ipynb: Jupyter notebook for training and evaluating the CLIP + CVAE model on the CelebA dataset.

你可以通过Kaggle下载CelebA数据集。以下是CelebA数据集的Kaggle下载链接：

CelebA Dataset on Kaggle

下载步骤：

1. 点击链接访问数据集页面。
2. 如果你还没有登录Kaggle账户，先登录你的账户。
3. 在页面上点击“Download”按钮下载数据集压缩包。

下载后，你可以解压文件，得到数据集的图像和标签文件。

注意：

你可能需要安装并配置Kaggle的API才能通过命令行进行下载。可以使用以下命令：

kaggle datasets download -d jessicali9530/celeba-dataset

确保你的Kaggle API密钥已经设置在环境变量中，具体的配置方法可以参考Kaggle API文档。