Hi friends.
I’m probably missing something really basic from the documentation and examples, but reading:
and
I could not grasp how does one update the step/stop parameter? I.e. how do I link training_iteration
ray expects to my model.
Furthermore, do I have any way of knowing if the scheduling/ optimization is “running”?
Meaning, when setting stop='foo'
(I have no foo
variable…) for tune.run()
, I get no warnings and the task runs transparently to the fact it does not know what is the training step - which means it cannot really early stop or optimize for better parameters.
Thanks!