Early stopping in NP training

[Amjad-AbuRmileh]

Is there an option to configure early stopping parameters for the neural prophet?

[ourownstory]

Hi @Amjad-AbuRmileh
If you want to speed up training, you can play with epochs, batch and lr, those are more effective with the one cycle policy than early stopping.

[Amjad-AbuRmileh]

I didn't understand the one-cycle policy thing?

[ourownstory]

we use this:

https://arxiv.org/abs/1803.09820

explanation:
https://towardsdatascience.com/finding-good-learning-rate-and-the-one-cycle-policy-7159fe1db5d6