The RandAugment Implementation Is Wrong

[psellcam]

google-research/fixmatch#65

As confirmed by the collaborator of the official implementation the randaugment function should be applied all the time not 50 percent of the time.

The only 50% chance from the original paper refers to the flips etc of weak augmentation only and definitely not the strong augmentation methods

This could explain why this repository does better than the original paper but it isn't what the original paper did.

For those who want the true performance of the original fixmatch paper you need to delete

if random.random() < 0.5:

From the https://github.com/kekmodel/FixMatch-pytorch/blob/master/dataset/randaugment.py file

[07Agarg]

Hi @psellcam , thanks for pointing out the issue.
I believe then this if random.random() < 0.5: is removed from the class RandAugmentPC and the class RandAugmentMC.

Let me know if I understand this correctly. Thanks again.

[psellcam]

Yes exactly!!

[kekmodel]

https://github.com/google-research/fixmatch/blob/d4985a158065947dba803e626ee9a6721709c570/third_party/auto_augment/augmentations.py#L213
and
https://github.com/google-research/fixmatch/blob/d4985a158065947dba803e626ee9a6721709c570/libml/augment.py#L198