Best way to train on all data without using validation set?

Part 1 (2019)
1

I have tuned my model and in the last step I want to train on all data including the validation set.
What is the best way to implement this? When I pass in None or an empty tensor as validation set to ImageDataBunch.create() I get errors. I vaguely remember there is a flag that can be set to use all data.

I am using fastai 1.0.24

2 Likes