Internal server error

Hello everyone,
First, I would like to thank the developers for DLS which finally allows me to play with Deep Learning.
I am a pathologist and my programming skills are pretty basic.
I started by testing MNIST which works perfectly on my config (compilation, training, test …)
I tried to reproduce a fairly classical Unet to segment cell nuclei. Unfortunately I can not start the training with either an empty error message or “internal server error” message
I think there must be an error in my pipeline but … what/where ???
Can you help me?
Thanks +++ :smiley:
PS .:
my dataset and pipeline can be found here
my dataset comes from there:
I just modified the initial images: I put them on a scale of gray and in png