How to configure when model training ends?

[adamkgoldfarb]

I searched pretty thoroughly through the docs for an answer, but couldn't find one-- apologies if this has already been asked:

How does spacy determine when training ends for a given session? Is there a space in the config where I can adjust this?

Thanks!

[polm]

Good question, it may have been hard to find this in the docs because there are multiple settings that control when training ends.

• max_epochs sets the number of epochs, typically complete passes over the training data, to use for training
• max_steps sets the max number of iterations, steps, or batches to process
• patience is an early stopping setting; training will stop after patience steps with no improvement

These are here in the docs. You can set all of these and the first one to be triggered will stop training. Typically you'd use one of max_epochs or max_steps with patience possibly triggering early stopping.

[adamkgoldfarb]

Incredibly helpful, thank you very much! These various settings make sense, and now it's clear how they interact with one another.