Skip to content

Conversation

@yaoyu-33
Copy link
Contributor

What does this PR do ?

Add a one line overview of what this PR aims to accomplish.

Changelog

  • Add specific line by line info of high level changes in this PR.

GitHub Actions CI

See the CI sectionin 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:

  • Make sure you read and followed Contributor guidelines
  • Did you write any new necessary tests?
  • Did you add or update any necessary documentation?
  • Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
    • Reviewer: Does the PR have correct import guards for all optional libraries?

If you haven't finished some of the above items you can still open "Draft" PR.

Additional Information

  • Related to # (issue)

@copy-pr-bot
Copy link

copy-pr-bot bot commented Dec 18, 2025

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@HollowMan6
Copy link
Contributor

HollowMan6 commented Dec 18, 2025

Update: now all are fixed

Update: Only linear_proj and linear_fc2 are correctly mapped with current version for LoRA:

image

Looks like those fused weights (e.g. fc1, qkv) in LoRA are still not mapped to hf_name correctly and they are still using the megatron naming:

image

Meanwhile, though Canonical LoRA is functioning in the sense of naming, it doesn't seem like it's working correctly, I will investigate this further:

Co-authored-by: ℍ𝕠𝕝𝕝𝕠𝕨 𝕄𝕒𝕟 <[email protected]>
Signed-off-by: Yu Yao <[email protected]>
@yaoyu-33
Copy link
Contributor Author

@HollowMan6 yes, I think I messed up a bit about the name conversion for fused base names. lemme try to fix

@HollowMan6
Copy link
Contributor

HollowMan6 commented Dec 21, 2025

@yaoyu-33 I've opened PR #1788 that targets to bridge/peft_bridge_1 branch for fixing expert layers' case, feel free to merge that one or integrate it manually into this PR.

The convergence situation is good on dense models for RL on verl, with the gray one representing Canonical LoRA with bridge, blue one representing the normal LoRA with bridge, and yellow one representing the LoRA merge.

image

The convergence tests for MoE (qwen3-30b-a3b):

image

HollowMan6 and others added 3 commits December 29, 2025 15:31
# Conflicts:
#	src/megatron/bridge/models/conversion/model_bridge.py
#	tests/unit_tests/models/test_model_bridge_lora.py
@yaoyu-33 yaoyu-33 changed the title [Draft] Peft Bridge Peft Bridge Jan 5, 2026
yaoyu-33 and others added 3 commits January 5, 2026 11:35
Co-authored-by: ℍ𝕠𝕝𝕝𝕠𝕨 𝕄𝕒𝕟 <[email protected]>
Signed-off-by: Yu Yao <[email protected]>
Signed-off-by: yaoyu-33 <[email protected]>
@yaoyu-33
Copy link
Contributor Author

yaoyu-33 commented Jan 5, 2026

/ok to test 4da571d

@yaoyu-33 yaoyu-33 enabled auto-merge (squash) January 12, 2026 19:23
@chtruong814
Copy link
Contributor

The code cov will be improved in follow-up PRs. Patch coverage in this PR is 78% and overall coverage is still 85%. Will go ahead and merge this.

@chtruong814 chtruong814 disabled auto-merge January 12, 2026 22:03
@chtruong814 chtruong814 merged commit 83a7c11 into main Jan 12, 2026
79 of 85 checks passed
@chtruong814 chtruong814 deleted the bridge/peft_bridge_1 branch January 12, 2026 22:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants