Skip to content

fix: bump mlm to fix NVBug 6432008 - #3265

Merged
terrykong merged 2 commits into
r0.7.0from
yukih/bump-mbridge-mlm
Jul 18, 2026
Merged

fix: bump mlm to fix NVBug 6432008#3265
terrykong merged 2 commits into
r0.7.0from
yukih/bump-mbridge-mlm

Conversation

@yuki-97

@yuki-97 yuki-97 commented Jul 17, 2026

Copy link
Copy Markdown
Contributor

cherry-pick NVIDIA/Megatron-LM#5715 in mlm based on current mlm commit.

…ocking

Signed-off-by: Yuki Huang <yukih@nvidia.com>
@yuki-97
yuki-97 requested review from a team as code owners July 17, 2026 12:32
@copy-pr-bot

copy-pr-bot Bot commented Jul 17, 2026

Copy link
Copy Markdown

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.

@github-actions

Copy link
Copy Markdown

✅ Submodule Fast-Forward Check Results

Check based on commit: a7e7afb (PR #3265 from yukih/bump-mbridge-mlm)

✅ Submodules that are properly updated:

Megatron-Bridge: ✅ PR branch is ahead of r0.7.0 branch (fast-forward)

All submodule changes look good! ✨

@yuki-97 yuki-97 added the CI:L1 Run doctests, unit tests, and functional tests label Jul 17, 2026
@qiaochuz-nv

Copy link
Copy Markdown
Contributor

Cherry-pick and verification completed.

Cherry-pick:
- PR: NVIDIA-NeMo/RL #3265, targets r0.7.0
- Created isolated worktree:
  /home/qiaochuz/nemo/RL_bv_6432008
- Branch: qiaochuz/nvbug-6432008-mlm
- Cherry-pick commit: eeb23dbab1fa6feb6480de1f72641c2f9afbdd6c
- Worktree is clean.

PR #3265 updates Megatron-Bridge to commit 151168f..., whose nested Megatron-LM is:
7b8495881dee1689db9d5ed636d8b6b48eacf260
- Includes Megatron-LM #5715.
- Verified transfer_bookkeeping_to_gpu() now uses:
  python
  self.gpu_view.buf.copy(self._cpu_bookkeeping_buf, non_blocking=False)


Fresh verification:
- Base container: nemo-rl-nightly-20260717.sqsh
- The released nightly itself remains unpatched, so the run used that nightly base with only the cherry-picked Megatron-LM source mounted over:
  /opt/nemo-rl/3rdparty/Megatron-Bridge-workspace/Megatron-Bridge/3rdparty/Megatron-LM
- Run ID: 20260717T171327Z-f03c198e
- SLURM job: 5659049
- Node: eos0044
- SLURM: COMPLETED 0:0, elapsed 00:03:58

Result:
text
test_grpo_megatron_generation: passed
effective exit code: 0
duration: 158s
metric: max(train/gen_kl_error) = 0.0004232723731547594 < 0.01

pass

@qiaochuz-nv qiaochuz-nv left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

run pass.

@terrykong
terrykong enabled auto-merge (squash) July 17, 2026 17:40
@terrykong
terrykong disabled auto-merge July 17, 2026 17:40
@terrykong
terrykong enabled auto-merge (squash) July 17, 2026 17:41
@terrykong

Copy link
Copy Markdown
Collaborator

/ok to test a7e7afb

@yuki-97

yuki-97 commented Jul 18, 2026

Copy link
Copy Markdown
Contributor Author

/ok to test 84d4921

@github-actions

Copy link
Copy Markdown

✅ Submodule Fast-Forward Check Results

Check based on commit: 84d4921 (PR #3265 from yukih/bump-mbridge-mlm)

✅ Submodules that are properly updated:

Megatron-Bridge: ✅ PR branch is ahead of r0.7.0 branch (fast-forward)

All submodule changes look good! ✨

@terrykong
terrykong merged commit 9d12b16 into r0.7.0 Jul 18, 2026
94 checks passed
@terrykong
terrykong deleted the yukih/bump-mbridge-mlm branch July 18, 2026 08:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CI:L1 Run doctests, unit tests, and functional tests

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants