Nan image

[gr8joo]

Hi, thank you for sharing your code!
I found that your code is easier to follow and intuitive!

One issue I have is I got the error like following when I execute 'python train-WGAN-GP-VDB.py'
.................................................................................................................................
Epoch: ( 0) ( 1/ 1875)
Traceback (most recent call last):
File "train-WGAN-GP-VDB.py", line 209, in
,alpha=Alpha,n_critic=n_critic,gpu_id=gpu_id,data_pool=data_pool)
File "train-WGAN-GP-VDB.py", line 176, in main
utils.imwrite(utils.immerge(f_sample_opt, 10, 10), '%s/Epoch_(%d)_(%dof%d).jpg' % (save_dir, epoch, it_epoch, batch_epoch))
File "/content/utils.py", line 363, in imwrite
return scipy.misc.imsave(path, to_range(image, 0, 255, np.uint8))
File "/content/utils.py", line 349, in to_range
assert np.min(images) >= -1.0 - 1e-5 and np.max(images) <= 1.0 + 1e-5 , 'The input images should be float64(32) and in the range of [-1.0, 1.0]!'
AssertionError: The input images should be float64(32) and in the range of [-1.0, 1.0]!
[*] MemoryData: stop threads and close session!
.................................................................................................................................

So I checked the value of image f_sample_opt in line 172 of train-WGAN-GP-VDB.py and found that the image was composed of all Nan values.

I found the same result with my local machine and Colab.

Do you have any idea why this happens?

Thank you.