Fix W&B step mismatch by consolidating log calls#15235
Open
yurekami wants to merge 1 commit intoNVIDIA-NeMo:mainfrom
Open
Fix W&B step mismatch by consolidating log calls#15235yurekami wants to merge 1 commit intoNVIDIA-NeMo:mainfrom
yurekami wants to merge 1 commit intoNVIDIA-NeMo:mainfrom
Conversation
The W&B _step counter was showing approximately 2x the actual
global_step because `log('global_step')` and `log('step')` were
called separately, causing WandbLogger's internal _step counter
to increment twice per training step.
This fix consolidates both metrics into a single `log_dict()` call
in all three strategy files, ensuring W&B step aligns correctly
with trainer.global_step for accurate cross-run comparisons.
Files modified:
- nemo/lightning/pytorch/strategies/megatron_strategy.py
- nemo/lightning/pytorch/strategies/fsdp_strategy.py
- nemo/lightning/pytorch/strategies/fsdp2_strategy.py
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: yurekami <yurekami@users.noreply.github.com>
yurekami
force-pushed
the
fix/wandb-step-mismatch
branch
from
b48eb36 to
5a02c95
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #15204
The W&B
_stepcounter was showing approximately 2x the actualglobal_stepbecause multiple separatelog()calls in the training step caused WandbLogger to increment its internal counter multiple times.Root cause: Each call to
self.lightning_module.log()triggers WandbLogger to auto-increment its_stepcounter. When loggingglobal_stepandstepseparately, the counter increments twice per training step.Fix: Consolidated the separate
log()calls into a singlelog_dict()call in:megatron_strategy.pyfsdp_strategy.pyfsdp2_strategy.pyTest plan
🤖 Generated with Claude Code