cp: fix(recipes): derive pretraining vocab from tokenizer (4996) into r0.6.0 - #5280
Merged
Merged
Conversation
Signed-off-by: Chen Cui <chcui@nvidia.com> (cherry picked from commit af17edf)
cuichenx
added
cherry-pick
breaking-change
cuichenx
added
area:recipe
cuichenx
added
full-test-suite
r0.6.0
4 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Backports #4996 to the 26.08 release branch.
The stock
nvcr.io/nvidian/nemo:26.08.rc3image contains the recipe and scheduler fixes, but its Megatron-Bridge checkout predates #4996. This cherry-pick makes pretraining derive vocabulary size from the tokenizer while preserving the explicit performance-recipe behavior.Validation:
af17edf52c514c58c79cd291b04a9b42bc5c57f32.18.0+e7c550c5