Is that ok to apply dropout when calculating the consistency loss ? - Githubissues

google-research / uda

Unsupervised Data Augmentation (UDA)

https://arxiv.org/abs/1904.12848

Apache License 2.0

2.17k stars 312 forks source link

Is that ok to apply dropout when calculating the consistency loss ? #79

Open shamanez opened 4 years ago