google-research / augmix

AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty

google-research/augmix Issues

torch version 1.2.0 not compatible with GPU version
Updated 8 months ago1
Nan loss for ResNext backbone trained on cifar 100
Updated 9 months ago1
none
Updated a year ago
Use combined with Object Detection
Updated 3 years ago
command for imagenet evaluation
Updated 3 years ago
Corruption acc. of a Resnet50 trained on cifar10 with augmix
Closed 3 years ago2
KL divergenve
Closed 3 years ago1
Jensen–Shannon divergence
Closed 3 years ago2
augmentations used in augmix
Closed 3 years ago3
What about the performance on regular classification problems?
Closed 3 years ago5
mean and std for every channel is 0.5 instead of (0.5071, 0.4867. 0.4408) as mean and (0.2657, 0.2565, 0.2761) as std
Closed 4 years ago2
Issue loading pretrained weights
Closed 4 years ago3
Repeated Evaluation in ImageNet
Closed 4 years ago1
`depth` is constant each `width`
Closed 4 years ago2
There is a big gap between the results of the code and the results in the paper on CIFAR10-C
Closed 4 years ago2
Testing with the best model
Closed 4 years ago4
A question in "third_party/WideResNet_pytorch/wideresnet.py out = self.relu2(self.bn2(self.conv1(out)))
Closed 5 years ago2
Reproduce of Baseline
Closed 5 years ago
Can you share any acc information per epoch?
Closed 5 years ago3
Little difference between with/without JSD loss
Closed 5 years ago1
CIFAR-10/ImageNet-P code
Closed 5 years ago2
The CIFAR-10-C dataset are not normalized
Closed 5 years ago1
ImageNet hparams
Closed 5 years ago3
I found an error
Closed 5 years ago2
About the loss function
Closed 5 years ago1