Why the lose value is Nan at the beginning of training sometimes？

nshaud / DeepHyperX

Deep learning toolbox based on PyTorch for hyperspectral data classification.

Other

439 stars 122 forks source link

Open Lzyin opened 3 years ago

Lzyin commented 3 years ago

Describe the bug A clear and concise description of what the bug is.

To Reproduce Steps to reproduce the behavior (e.g. the command that you used).

Expected behavior A clear and concise description of what you expected to happen.

Desktop (please complete the following information):

nshaud commented 2 years ago

With what parameters? Can you give some more details to reproduce this problem?