Possible BUG, regarding the self.epoch parameter

[zhangyikaii]

Hi,
I found that self.epoch may not be updated, which causes the _update_weight function to not work.
Am I missing something? How can I reproduce the experiment correctly?

Thank you very much for taking the time to help.

[zhangyikaii]

A brief additional note:
The parameter self.epoch is used here, but it is not updated anywhere else. I am trying to update self.epoch within the training loop (here) to reproduce the original result.
Is there any problem with this? Thanks a lot.