Welcome to this GAN repository where we venture into the mesmerizing world of Generative Adversarial Networks. If you're passionate about GANs, this is the place for you!
The repository is organized into two main sections:
-
DCGAN:
- model.py: Contains the implementation of the DCGAN model inspired by the paper "Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks". Here you'll find the architectures of both the Discriminator and Generator.
- train.py: This is where the magic happens! We train our DCGAN on the MNIST dataset. Everything from setting up hyperparameters to initializing data loaders is here.
-
simple_gan:
- simple_gan.py: Looking for something a bit simpler to start with? This file contains a basic GAN implementation using linear layers. It's a great entry point if you're new to GANs. Plus, there are some handy notes on potential improvements and tweaks!
-
Clone the repository:
git clone <your-repo-link> -
Navigate to the DCGAN folder:
cd DCGAN -
Train the DCGAN model:
python train.py -
Or, if you're interested in the simpler GAN, navigate to the simple_gan folder:
cd simple_gan -
And train the simple GAN:
python simple_gan.py
If you have ideas, improvements, or have found bugs, feel free to open an issue or submit a pull request. All contributions are welcome!
Thanks for stopping by! Dive in, experiment, and have fun with GANs. And remember, the world of deep learning is vast and full of possibilities, so keep exploring!