Implementation of a WGAN for MNIST, with checkpoint in model/
- tensorflow=1.0.0
- numpy
- tqdm
- argparse
To train the model, run the command:
python model.py --train
Model retrieve automaticcaly the last checkpoint and restart at the global step, and saved itself every 100 iterations.
It also saved summary for images generated and other variables, to further
display in Tensorboard tensorboard --logdir="logs"
To draw sample from the generator, run the command:
python model.py --draw 1,2,3,4,5,7,9
