Hi!
Sometimes, while I’m training a model, my jupyter notebook disconnects. When I reconnect, I cannot say updates in the training, but the kernel is still running, this is, the model is still being trained.
I’d like to wait so I don’t lose my progress, but, how can I report my error metrics after training the model?
So after this:
learn.fit_one_cycle(10)
I was thinking of some methods like:
learn.trains_loss()
learn.valid_loss()
etc.
Is this possible? If so, how?
Thanks!