How to add break point for training an ensemble classifier?

Hi,

I am running RUSBoost esemble classifier on a cluster which has time limitation for 7 days, and it will stop training in midway. I do not know how to tackle my issue. Is there any way to have a break point for training of fitcensemble function, which it can be able to save the model after some duration and continue training from that point again (similar strategy like snapshotting in deep learning)?

Or is it possible to set the ensemble classifier to run on GPU or parallelly on CPU?

You opinion is really appreciated.