README.md

Simple GAN Implementation using Tensorflow 2.0

This is a Generative Adversarial Network (GAN) implementation for MNIST image generation. We can create a vanilla GAN and conditional GAN in about 60 lines of Tensorflow 2.0 code.

Requirements

• Python 3
• Tensorflow 2.0
• CUDA 10.0 or later
• cuDNN 7.4.1 or later
• If your system doesn't meet these requirements, you can use Google Colaboratory (tutorial).

Results

• Vanilla GAN

• Conditional GAN

References

• Tensorflow 2.0
• Vanilla GAN paper
• Conditional GAN paper