|
|
$ git clone https://github.com/Bex0n/CelebA-VAE
$ cd CelebA-VAE
$ pip install -r requirements.txt
Download the dataset link and extract:
unzip celeba.zip
cd celeba
unzip img_align_celeba.zip
cd ..
python train.py -c <path_to_config_file>
If you want to train the model with cpu only, change accelerator: 'auto' to accelerator: 'cpu' in the provided config file.
Note: model capabilities such as generating new celebrities or modifying features are available in the file experiments/vanilla_vae.py