I looked into the code and found that functional.batch_norm() added a check ... value per channel when training, got input size {}'.format(size)) f = torch.
確定! 回上一頁