Fix NameError in pretrain_retro.py (add import_module), remove unused… #2084
Conversation
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
… import, and clarify log message
I fixed a runtime error and cleaned up minor issues in pretrain_retro.py:
What I changed
Added missing import:
from importlib import import_module Reason: import_module(...) is called in core_model_provider when args.spec is set; without this import the code throws a NameError at runtime.
Removed unused import:
MegatronTokenizer (unused) — removed to avoid noise and potential lint warnings.
Improved log message:
Changed print_rank_0('building GPT model ...') → print_rank_0('building Retro model ...') to accurately reflect the model being constructed.
Why
The missing import is a definite runtime bug that will break the experimental spec-loading path (args.spec).
Removing unused imports helps keep the codebase clean and avoids linter complaints.
Updating the log message reduces confusion when reading logs (this script builds a Retro model, not GPT).
Testing / How I validated
Static review to confirm import_module usage and the missing import.
Confirmed MegatronTokenizer is not referenced elsewhere in the file.
Confirmed the corrected log message matches the behavior of core_model_provider.
Risk / Backward compatibility
Low risk: only adds an import, removes an unused import, and updates a log string. No behavior/functionality changes to model logic.
If other modules relied on MegatronTokenizer being imported at this module level (unlikely), they should import it explicitly from megatron.core.tokenizers.
yashaswikarnati
approved these changes
Nov 2, 2025
Contributor
|
/ok to test d019406 |
Contributor
|
Thank you for your contribution! NVIDIA Megatron-LM is currently transitioning to development on Github. We will aim to review your PR after we complete our transition and stabilize our Github development process. Thank you for your understanding. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
I fixed a runtime error and cleaned up minor issues in pretrain_retro.py:
What I changed
Added missing import:
from importlib import import_module Reason: import_module(...) is called in core_model_provider when args.spec is set; without this import the code throws a NameError at runtime.
Removed unused import:
MegatronTokenizer (unused) — removed to avoid noise and potential lint warnings.
Improved log message:
Changed print_rank_0('building GPT model ...') → print_rank_0('building Retro model ...') to accurately reflect the model being constructed.
Why
The missing import is a definite runtime bug that will break the experimental spec-loading path (args.spec).
Removing unused imports helps keep the codebase clean and avoids linter complaints.
Updating the log message reduces confusion when reading logs (this script builds a Retro model, not GPT).
Testing / How I validated
Static review to confirm import_module usage and the missing import.
Confirmed MegatronTokenizer is not referenced elsewhere in the file.
Confirmed the corrected log message matches the behavior of core_model_provider.
Risk / Backward compatibility
Low risk: only adds an import, removes an unused import, and updates a log string. No behavior/functionality changes to model logic.
If other modules relied on MegatronTokenizer being imported at this module level (unlikely), we should import it explicitly from megatron.core.tokenizers.