How to add validation dataframe in Tabular databunch, without giving split_by

Part 1 (2019) Advanced (Part 1 v3)
1

I want to add a custom validation dataframe to my Tabular databunch. Hence, I don’t want to use a random split or any of the methods provided in the API. How can I do that?