Halting training

[Lawrence]

Is there any built in functionality for halting training when "error" stops decreasing?

Or do we need to monitor that ourselves?

[andrew.kirillov]

It is you who is driving the process. You call Run() to do training iteration. Then you check whatever you need to check (error, number of iterations, etc.) and decide if you want to go further.

[Lawrence]

Understood - thank you.