How to test?

[binwang777]

Excellent work, thanks for your upload.
I know that it is progressive during training. So how do you save your parameters, is it to save the parameters of this entire process, or only the parameters of the last pass through the model. Similarly, if the parameters of the entire process are saved, test how to proceed.

[yangyixiao]

Hi
In figure.3 in our paper, all G_T indicate a same network which means they have the same weights. The weights in one iteration won't change until we update them with Adam optimizer.