Can I stop the experiment when I see good results?


I understand that the running time is that costs money. Do I have to wait until the experiment stops itself or can I decide to stop the experiment when I see that the experiment precision and accuracy are good? Because what I get as lower value is just an asymptotic/tangential like curves.

Any advice?

There’s a feature activated by default on the modelling view called Early Stopping. The purpose with that setting is that it will automatically stop the training if the results aren’t getting any better.

This is what you usually want. If you, for another reason, want to stop training, you can do that by pressing the pause button on the experiment.

You can read more about it here

1 Like