How to know that my Siamese Network is doing well?

[Youyou]

Hi guys, i have a simple question actually. i'm training a siamese network, so as output i have something similar to the siamese_mnist example: I have the loss and the learning rate. But with just those informations, how could i know if my Network is doing well?. Normally for the other tutorials there was an accuracy layer that outputs a percentage of the right predictions. But with the siamese network i have two final output from my two data layers(in fact it is one data layer sliced) feat and feat_p that feeds a contractive loss function.

I don't know how to implement an accuracy layer with those kind of parameters. So do anyone know how can i see if my network is doing a good job or not? 

[Arash]

Hi, this is also my issue. Have you found the answer?