[mxfp8 moe training] register constant with pytree#3667
Merged
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3667
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New Failures, 4 Unrelated FailuresAs of commit 8a86e54 with merge base 3834780 ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
liangel-02
approved these changes
Jan 21, 2026
Contributor
Author
|
test failures unrelated, landing |
jcaip
pushed a commit
that referenced
this pull request
Jan 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
After #3606
@torch._dynamo.nonstrict_tracewas added to support autograd functions having different tensor subclasses for forward output and backward input.The e2e integration tests for compile and eager passed and it was landed, but it turns out if you explicitly specify a scaling recipe instead of defaulting, it breaks with this error:
Basically we need to register the enum with pytree, which is done in this PR.
Tests
Added unit test cases which fail without this change.
pytest test/prototype/moe_training/test_scaled_grouped_mm.py -k dq