fix: correct mcore dtype + assertion on activation_func #572
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
Signed-off-by: Terry Kong <terryk@nvidia.com>
parthchadha
approved these changes
Jun 27, 2025
therealnaveenkamal
pushed a commit
to therealnaveenkamal/RL
that referenced
this pull request
Jul 7, 2025
KiddoZhu
pushed a commit
that referenced
this pull request
Jul 28, 2025
Signed-off-by: Terry Kong <terryk@nvidia.com>
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Outdated suggestions cannot be applied.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.
This is to address a bug found when testing Nemotron-H where the params_dtype gets set incorrectly and add an assert to give a hint of what might have happened if activation_func is empty.
The issue was during conversion we use the default value of the optimizer.params_dtype as float32 here https://github.com/NVIDIA/NeMo/blob/bab66472d2f2eb05ab621dbad66ad6031e4ee19e/nemo/tron/converter/common.py#L217. But the params_dtype is supposed to be set by the
bf16andfp16args according to how it's handled in mcore: https://github.com/NVIDIA/Megatron-LM/blob/1ab876ddc4c1893c76f26d775226a8d1dcdfb3d2/megatron/training/arguments.py#L676This change just respects that logic.