Sungjin Kim

About julia files in examples - overfitting

2

I found that 'test-wasserstein.jl' does not consider overfitting yet although it includes AccuracyLayer. If my understanding is right, I am suggesting to include a new extended example. In neural networks,...

For DenseNet, Convolution2D is replaced by Conv2D. For testing, Py2 is changed to Py3. I tested after the updations and it works very fine.

I updated the original code to support Keras2. Moreover, MNIST dataset can be used without additional modification except for the flag, 'load_data_name'. It can be one of 'mnist' and 'cifar10'.

print() included

Update example_gan.py

1

Code for dividing an integer to get an integer result is changed to work appropriately. E.g., hidden_dim / 4 --> int(hidden_dim / 4)