Cherry-pick: Fix mtp_num_layers and clip_qk issues (#2581, #2776)

[BestJuly]

What does this PR do ?

Cherry-pick two merged PRs from dev branch to main.

The followings are from original PR description.

PR#2581 Fix: don't enter branch if mtp_num_layers == 0

The condition if self.config.mtp_num_layers is not None evaluates to True when mtp_num_layers=0, causing the code to enter the MTP branch and crash on labels.clone() when labels=None.

Since range(0) produces an empty loop and torch.chunk(..., 1) is essentially an identity operation, entering this branch with mtp_num_layers=0 does nothing useful but requires labels to be passed.

PR#2776 Fix clip_qk for virtual pipeline size > 1

Add a None check for current_max_attn_logits in clip_qk.

When virtual pipeline is enabled with size > 1, in the first iteration step, only some model chunks have initialized the current_max_attn_logits, other chunks remain with current_max_attn_logits = None, which would lead to a TypeError when all_reduce is called in the following step.

⚠️ For major changes (either in lines of code or in its impact), please make sure to first share a design doc with the team. If you're unsure what's the best way to do so, contact the @mcore-oncall.

Contribution process

flowchart LR
    A[Pre-checks] --> B[PR Tests]
    subgraph Code Review/Approval
        C1[Expert Review] --> C2[Final Review]
    end
    B --> C1
    C2 --> D[Merge]

Loading

Pre-checks

• I want this PR in a versioned release and have added the appropriate Milestone (e.g., Core 0.8)
• I have added relevant unit tests
• I have added relevant functional tests
• I have added proper typing to my code Typing guidelines
• I have added relevant documentation
• I have run the autoformatter.sh on my PR

Code review

The following process is enforced via the CODEOWNERS file for changes into megatron/core. For changes outside of megatron/core, it is up to the PR author whether or not to tag the Final Reviewer team.

For MRs into `main` branch

Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!

(Step 1): Add PR label Expert Review

(Step 2): Collect the expert reviewers reviews

1. Attach the Expert Review label when your PR is ready for review.
2. GitHub auto-assigns expert reviewers based on your changes. They will get notified and pick up your PR soon.

⚠️ Only proceed to the next step once all reviewers have approved, merge-conflict are resolved and the CI is passing.
Final Review might get declined if these requirements are not fulfilled.

(Step 3): Final Review

1. Add Final Review label
2. GitHub auto-assigns final reviewers based on your changes. They will get notified and pick up your PR soon.

(Optional Step 4): Cherry-pick into release branch

If this PR also needs to be merged into core_r* release branches, after this PR has been merged, select Cherry-pick to open a new PR into the release branch.

For MRs into `dev` branch

The proposed review process for `dev` branch is under active discussion.

MRs are mergable after one approval by either eharper@nvidia.com or zijiey@nvidia.com.

Merging your PR

Any member of core-adlr and core-nemo will be able to merge your PR.

[maanug-nv]

Since these are both bug fixes, it seems a good idea to add unit tests to this PR to ensure same things don't happen again. For example, test model creation with mtp_num_layers=0 doesn't crash.

[jaredcasper]

Please use descriptive names for PRs, not referring to PR #s.