Simple GANs example code
All codes running in Python3.
You have to check the packages first, and use pip install
to install if you miss them:
- Tensorflow==1.3.0
- Numpy
- Scipy
- urllib
- tarfile
- pickle
- six
Run
python 01-cifar10_DCGAN_slim.py --bs 256 --dlr 0.0002 --glr 0.0002
Results & Models
The training results are saved in the [result]
folder.
The tensorflow checkpoint models are saved in the __models__
folder.
Visualization
You can use tensorboard
to see the training progress online.
First, toggle the port:
$ tensorboard --logdir=summary/
Tensorboard 0.1.5 at http://ServerName:6006 (Press CTRL+C to quit)
Then open your browser, go to http://localhost:6006
. You can see the histogram of inception score progressing.
Examples of generated images
- mnist
- cifar10
- imagenet