Unable to reproduce Cifar-10 results for WRN-28-10

ildoonet / pytorch-randaugment

Unofficial PyTorch Reimplementation of RandAugment.

MIT License

627 stars 98 forks source link

Unable to reproduce Cifar-10 results for WRN-28-10 #32

Open humza909 opened 1 year ago

humza909 commented 1 year ago

Hi,

I am unable to reproduce cifar-10 results for the model WRN-28-10. The accuracy I get stays as low as 96.7% but the reported accuracy is 97.4% (this repo) / 97.3% (in paper). Anyone else facing similar issues? @ildoonet Is your reported accuracy the average of multiple runs or only for a single run?

I am using the following command:

python RandAugment/train.py -c confs/wresnet28x10_cifar10_b256.yaml

Thanks

dahuaxiya commented 1 year ago

I get 96.7% test accuracy too after 200 epochs. Do you know why now?