Testing Image Problem

Hello,
Does anyone know, how separate the image for training and testing? After training, if I use the same folder with same training dataset (celebA/mnist/different) for "Testing Mode" it can generate the images. But when I try to use different image for "Testing Mode" it shows error whether checkpoint error / there is no attribute as saver for DCGAN.

Any kind of help will be really appreciable.
Thanks
Hosna

No file, test, entry point, or complete error message is provided. Start by reproducing Testing Mode with a different image folder and capture the checkpoint or missing-saver error; the issue is done only when separate training and testing datasets work without that error.

Written by the indexing model from the issue text.