KaltsitI

The problem of reproducing the result of ImageNet

4

Recently I try reproducing the result in the paper. I successfully did this on cifar10 and GAN, but the test accuracy on ImageNet is nearly 69.5%, which on the paper...

I use acprop to train transformer, but it breaks down and keeps 0 bleu. Are there some bugs in the code? the transformer is from https://github.com/juntang-zhuang/fairseq-adabelief lr=5e-4 and 2e-5, eps=1e-16