I want to get the prob the input two picture?

[mgq1507]

If I change the loss='binary crossentropy' into loss='softmax',I can get the prob?
If you can give me some advice? I'm appreciate it.
Another, I have trained the network but it costs too much time,how long you train?

[mgq1507]

@anson0910

[anson0910]

I think you should be able to get the prob similarly by
prob = self.model.predict([img1, img2])

I recall using GTX 970, it took me around 8 hours to train.

[mgq1507]

Thank you ,my traing loss curve like this:

The curve not like yours.
s_net.train(starting_batch=40000),The trained model store where.
In the following code,the test example, not like yours.
Can you tell me how I can recurrent your result?
Thank you again,@anson0910.

[anson0910]

I'm not sure why your results are quite different..
Regarding how to save the model, you can check this link.

Hope it helps, thanks!