Face-Generation-GAN

I build a generative adversarial (GAN) network to generate new images of faces.

I first test the GAN on the MNIST dataset to generate new handwritten digits. Then I train the GAN on the CelebA dataset for generating new human faces.

The GAN is built on TensorFlow, written in Python 3 and is presented via Jupyter Notebook. Trained via FloydHub gpu instance.