Any changes to the function "optim.AdamW8bit"?

[arceus-jia]

When I use version 0.37.0 of AdamW8bit as an optimizer, the loss calculation quickly becomes nan, (which is problematic when training dreambooth or Lora) but when I fall back to version 0.35.4, these problems will no longer exist

[ArrowM]

Bump. (plus I think this is a dup of #121). Something changed between 0.35.0 and 0.36.0.

[ArrowM]

This is a total shot in the dark, but I wonder if bitsandbytes/functional.py:218-223 was accidentally indented in 2f2063b

[rationalism]

@ArrowM @arceus-jia This should be fixed by 0.41.1

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.