Nan appears during training
HospitableHost opened this issue · 1 comments
HospitableHost commented
garvita-tiwari commented
We figured that this occurs due to weight norm layer of pytorch: pytorch/pytorch#19126
Train without weight normalization
HospitableHost opened this issue · 1 comments
We figured that this occurs due to weight norm layer of pytorch: pytorch/pytorch#19126
Train without weight normalization