[model] feat: Add limited Gemma4 dense model support#3885
Merged
Conversation
yaoyu-33
previously approved these changes
May 20, 2026
Contributor
|
/claude review |
Contributor
Light Review - Issues found: (1) Typo gemma4_provider.py:349 decided base on should be decided based on (2) Missing space in error message gemma4_vl_bridge.py:83 f-string period has no trailing space so text runs together (3) Confusing error wording gemma4_vl_bridge.py:81 or model without hidden_size_per_layer_input is hard to parse (4) Copy-paste docstring test_gemma4_vl_conversion.py:300 says MoE bridge should say Dense bridge. See inline comments for code suggestions. Suggested test cases: No perf tests impacted. |
Contributor
|
@pavelgein Please address Claude's comments. Thanks |
Contributor
|
/ok to test f6f5420 |
Contributor
|
@pavelgein Please address the CI failures. I will run the CI again once addressed. |
Contributor
|
/ok to test 8ea504c |
kamran-nvidia
previously approved these changes
May 21, 2026
Signed-off-by: Pavel Gein <pavel.gein@gmail.com>
Signed-off-by: Pavel Gein <pavel.gein@gmail.com>
Signed-off-by: Pavel Gein <pavel.gein@gmail.com>
Signed-off-by: Pavel Gein <pavel.gein@gmail.com>
Contributor
Author
|
I have added support for dense Gemma4 model (not VLM one), so could you please review once again? |
yaoyu-33
approved these changes
May 22, 2026
Contributor
|
/ok to test 2305b42 |
vasunvidia
pushed a commit
to vasunvidia/Megatron-Bridge
that referenced
this pull request
Jun 10, 2026
Signed-off-by: Pavel Gein <pavel.gein@gmail.com> Co-authored-by: Kamran Jafari <kjafarisadeg@nvidia.com> Signed-off-by: Vasudevan Rengasamy <vrengasamy@nvidia.com>
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
4 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.
What does this PR do ?
Current implementation of Gemma4VL Bridge supports only MoE models.
This restrictions is loosed to support Gemma4 31B model, since it does not have a per-layer embeddings
Changelog
GitHub Actions CI
See the CI section in the Contributing doc for how to trigger the CI. A Nvidia developer will need to approve and trigger the CI for external contributors.
Before your PR is "Ready for review"
Pre checks:
If you haven't finished some of the above items you can still open "Draft" PR.
Additional Information