PyTorch implementation of Generative Adversarial Networks by Ian Goodfellow.
Please feel free to play around with the code. The aim of this repository is to allow researchers to try different ideas in regards to GANs.
-
Create your anaconda env
conda create -n myenv python=3.7
-
Install packages using pip
pip install -r requirements.txt
python main.py --n_batch 64 --seed 10You can take a look at the flags for any other hyperparameters you would like to experiment with.
GANs are notoriously hard to train. They are extremely sensitive to hyperparameters, especially the learning rate. However, you can try tinkering around with the learning rate to achieve better results.
random noise ❌
Please feel free to contribute. Pull requests are welcome. My goal is to manage a repositor that is comprehensive in nature, and easy to understand for beginners.