Using batch-normalization and dropout with Fast.ai library

Part 1 (2018) Beginner (2018)
1

I have worked through Jeremy’s dogs vs cats notebook but I am unable to find how to apply dropout and batch normalization with learn = ConvLearner.pretrained(arch, data, precompute = True)

I have used the corresponding functions in Tensorflow and Keras but I want to use fast.ai library for a Kaggle Competition.

Thank you very much for the help!

2

You’ll need to disable precomputation. Then you get dropout and batchnorm for free. Search the forums for details on setting the dropout rate, or watch the lessons where we cover that.

2 Likes