About code availability and experimental comparisons.

[Godforever]

I have read the paper on Pure Noise to the Rescue of Insufficient Data and found it very simple and exciting. However, I have the following questions:

1. When will the code be available?
2. Are the experimental comparisons fair? The authors use different network architectures on CIFAR-10/-100 and ImageNet, compared to the commonly-used setting.
3. Have the authors tried to experiment with ResNet-32 on CIFAR-10/-100？ I tried reproducing the code but didn't get better results than MiSLAS.

[opooladz]

can u post ur code for it? Also trying to re-produce rn. did you implement the updated version of BN?

[n12iki]

@Godforever and @opooladz did either of you get it working?