@crowsonkb I am not convinced that this is actually necessary. The implementation of the original authors does not do this. In fact, it seems to me as if they explicitly changed their implementation to undo the changes in your PR. Notice also that we do use the partial term in our final update. Can you provide more details on whether and if so why you still think this is useful?

Hello, I'm a bot! I can make mistakes, notify gabrieldemarmiesse if I made one.

I see that you modified the following file:

• keras_contrib/optimizers/padam.py

The owner of those file is @MFreidank

@MFreidank could you please take a look at it whenever
you have the time and add a review? Thank you in advance for the help.

@MFreidank please ignore the bot's message, my bot is up and running and it was just triggered for this PR.

@gabrieldemarmiesse
This has been stale since my comment from three months ago, so I suggest closing it unless you find a flaw with my arguments above.