Training is always an adventure—there are so many pitfalls when developing GAN architectures. In this training class, we aim to provide a simple setup to train a GAN that takes a 2D image and creates a 3D model.
This is the final recipe in our chapter, so we've got a few files to create—the train.py
, run.py
, and run.sh
files. Before continuing, check to make sure you have the exact same directory structure in your directory:
├── data ├── docker │ ├── build.sh │ ├── clean.sh │ ├── Dockerfile │ └── kaggle.json ├── out ├── README.md ├── run_autoencoder.sh ├── run.sh └── src ├── discriminator.py ├── encoder_model.h5 ├── encoder.py ├── gan.py ├── generator.py ├── run.py ├── train.py ├── x_test_encoded.npy └── x_train_encoded.npy