/pytorch-wgan-gp

PyTorch implementation of "Improved Training of Wasserstein GANs", arxiv:1704.00028

Primary LanguagePythonMIT LicenseMIT

pytorch-wgan-gp

PyTorch implementation of Improved Training of Wasserstein GANs, arxiv:1704.00028

loss-function-with-penalty

Results

Generated samples after training 1 epoch on LSUN Bedroom dataset

generated samples

Installation

$ git clone https://github.com/kuc2477/pytorch-wgan-gp && cd pytorch-wgan-gp
$ pip install -r requirements.txt

CLI

Train

$ # To download LSUN dataset (optional)
$ ./lsun.py --category=bedroom          

$ # To Run a Visdom server and start training on LSUN dataset.
$ python -m visdom.server
$ ./main.py --train --dataset=lsun [--resume]

Test

$ # checkout "./samples" directory
$ ./main.py --test --dataset=lsun

References

Author

Ha Junsoo / @kuc2477 / MIT License