Thanks for looking into this. Please note I am not using the latest tensorflow-macos and tensorflow-metal. I used versions informed by others that work.
I never try TF on a Mac till now, so I don't have any comparison with prior version to know if this is a regression.
This behaviour suggests a workaround where I just have to ensure every batch has the same max_len and padding, which may preclude certain memory and performance saving technique. This is not repro on Google Colab (cuda with T4)
kechan
Users receive reputation points by contributing quality content to the forums.
Posts include post and replies published on the forums.
Post authors can mark replies as Accepted to indicate successful solutions.
Apple Developer Forums admins can mark replies as Apple Recommended to indicate an approved solution
Last seen
User for