cifar_unnormalized should have normalize=False

brain-research / realistic-ssl-evaluation

Open source release of the evaluation benchmark suite described in "Realistic Evaluation of Deep Semi-Supervised Learning Algorithms"

Apache License 2.0

458 stars 98 forks source link

Closed joschu closed 5 years ago

joschu commented 5 years ago

build_tfrecords.py appears to have a bug where normalize=True is passed in for both cifar10 and cifar_unnormalized.

craffel commented 5 years ago

Thanks. We don't actually use cifar_unnormalized anywhere though, it should probably just be removed entirely.