Too large step_size in deepedit

[KumoLiu]

Describe the bug
Too large step_size which may cause the learning rate never decreases unless the model is trained for 5000 epochs.

tutorials/deepedit/ignite/train.py

Line 268 in bbc4e18

lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=5000, gamma=0.1)

Expected behavior
Set smaller step_size or set epoch_level=False in LrScheduleHandler

Originally posted in Project-MONAI/MONAI#6625

[KumoLiu]

Hi @diazandr3s, could you please help take a look at this issue?
Thanks in advance!

[diazandr3s]

Hi @KumoLiu,

Thanks for the ping. Good point!
Have you tried reducing the step_size? you may also want to consider changing the learning rate and/or optimizer.
As you may know, intensity normalization may also affect the training.

[KumoLiu]

Hi @diazandr3s, I didn't try to reduce the step_size, I'm assuming there's a typo here because the value of 5000 is set too high if it's epoch-level.
And if it's a typo, I can submit a PR to fix it.

Thanks!