lr_scheduler Exponential decay bug

[JachinJiang]

in lr_scheduler.py class ExponentialLr forget self.set_lr(self.current_lr);

[JachinJiang]

and when I reload model, in data_loader.py, __balance_val_split will be random
train_test_split(
np.arange(targets.shape[0]),
test_size=val_split,
stratify=targets,
random_state = 0. # add
) will resolve

[zhutmost]

in lr_scheduler.py class ExponentialLr forget self.set_lr(self.current_lr);

Hi thanks. Could u provide a pull request?

[JachinJiang]

Thanks for your code. I have pulled.