Update tests with zero3 for RLOO and GRPO as xfail only with transformers >= v5 - #5420
Merged
Merged
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
qgallouedec
approved these changes
Apr 2, 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Update tests with zero3 for RLOO and GRPO as xfail only with transformers >= v5.
This PR updates the test suite to more accurately handle failures related to ZeRO-3 when using newer versions of the
transformerslibrary. Specifically, it refines the test expectations so that tests using ZeRO-3 are only expected to fail whentransformersversion is 5.0.0 or higher.I tested locally that the error appeared in transformers v5.0.0 after the merge of:
See investigation in underlying issue:
Changes
Test suite improvements:
test_rewardandtest_rlootest cases intests/distributed/test_distributed.pyto mark ZeRO-3 tests as expected to fail only whentransformersversion is 5.0.0 or newer, instead of unconditionally. This makes the test results more accurate and prevents false negatives on older versions.Note
Low Risk
Low risk: only adjusts pytest
xfailconditions for distributed tests, affecting test expectations rather than runtime code paths.Overview
Updates distributed test parametrizations for
test_rlooandtest_grposo thezero3configuration is markedxfailonly whentransformers >= 5.0.0(withstrict=True), instead of always expecting failure. This reduces false failures on oldertransformersversions while retaining the known ZeRO-3 regression coverage.Written by Cursor Bugbot for commit 80d5dd6. This will update automatically on new commits. Configure here.