I'm having the same issue. I didn't notice the problem until today. I spent the last hours figuring out why my data is not syncing until I found this post. I'm using the same approach, restoring Core Data from a big JSON file with a lot of many-to-many relationships. At least now I know that the problem is not on my side...

Just tried with iPadOS 15.2 (beta 2) and Xcode 13.2 beta (13C5066c). Seems to be resolved!

Same problem here. Training worked fine 4 weeks ago. Since then I upgraded to macOS Monterey and the same script crashes using Adam, whereas it starts with SGD as optimizer. MacBook Pro (15-inch, 2017), Radeon Pro 560 4 GB as well as Radeon Pro 580 16GB as eGPU. I already have reinstalled tensorflow-metal and tensorflow-macos but it doesn't help.

Thank you! This was the only way that worked out for me...

Any news on that? We just got the error after releasing an update to production! We haven't had any issue in TestFlight migrating from one model version to the next, but now users are reporting that their data is gone with the same error messages. Constraint unique violation: Cannot migrate store in-place: constraint violation during attempted migration, UNIQUE constraint failed: ANSCKRECORDMETADATA.ZENTITYID, ANSCKRECORDMETADATA.ZENTITYPK.

Thank you so much! This really seems to work for now. I've been looking for days to get around this problem...