Fastai v2 "TTA" is not working in google Colab

[hagarciapinto (Henry)]

I am trying to use fastai v2 tta in google colab. I used the following command:

predictions, targets = learn.tta(dl=dl_test)

And I get the following error message:

AttributeError: ‘NBProgressBar’ object has no attribute ‘wait_for’

Any suggestions?
Thank you,
Henry

[mrfabulous1]

Hi hagarciapinto I hope you are having a wonderful day!

Maybe the posts in this thread can help you solve your error.

Fastai2 Test Time Augmentation

Hi all, I’m trying to understand how test time augmentation is implemented in fastai2. In fastai1, the default tta looks like it had 4 corner crops along with a horizontal flip + some other things In fastai2, what data augmentations does FA2 use? Are they set to the validation dataloader settings by default? I was initially looking for a the basic 4 corners + horizontal flip. I saw this in the FA1 source code, but did not see it in FA2. source code. Daniel Lam

This one looks like it may do it.

Fastai2 Test Time Augmentation

Just letting everyone know that on the issue associated with this on Github a workaround was identified and a pull request has been submitted. The workaround is set the learner epoch to zero right before calling tta. For example learn.epoch = 0 preds = learn.tta(dl=tst_dl, n=5) Updates on Github at: Fastai2 Test Time Augmentation

Cheers mrfabuluous1

[hagarciapinto (Henry)]

Thank you @mrfabulous1 , much appreciated.