How many pieces of data do you have for training? How many layers do you use in your model? How many epoches do you train to reach the best point usually?

This sounds like a stupid question. It is depends on data and model complexity. But I guess we may find some regular pattern from your answer. Do you want to find it out by saying your number?