How to handle imbalanced dataset in PyTorch?

Question

all.

I have some problems when I handled an imbalanced dataset.

I tried to use WeightRandomSampler to over-sampling/under-sampling, but the result is worse tha