training procedure

[zxnga]

Hey could you explain how to train your solution in order to reproduce the results please, I'm interested in your approach my objective being applying the solution on my own dataset.
Thanks a lot !