Gradients are zero after model is restored

[ameliajimenez]

Hello,

Someone else also has the problem that the gradients are initialized to zero after restoring the model? I would like to instead load the weights from the last saved point of training? Can someone help please?

Best regards, Amelia.

[jakeret]

How are you reloading the model?

Is the prediction working when using the predict function of the model?

[ameliajimenez]

Hi, thanks for the answer!

I haven't modified the code, so the model is restored in the same way as the code here.

Yes, prediction is working.

[jakeret]

When creating a new instance of Unet the global graph is resettet using tf.reset_default_graph() to avoid problems with previous graphs. Not sure if this is causing the issue you're seeing