Split on retain/forget set

[Yiwei Fu]

Hi, I have a problem regarding split on retain/forget set. Is there a fixed percentage? Since if the forget set is very large, it would make more sense to just train from scratch on the retain set.

[etriantafillou]

Hi,

Thank you for the question!

Our competition is now open for submissions on Kaggle, where the setup we are using is a little different than the starting kit.

For more info, see our latest email announcing the launch, the Kaggle page, and the document we put together describing the setup, including info on the retain / forget split.

Please let us know if you have any further questions!

Thanks,

Eleni