[Dev] Changes to support multimodule pipelining

[yashaswikarnati]

What does this PR do ?

Changes in 1F1B schedule to support multi module pipeline communication and unit tests

⚠️ For major changes (either in lines of code or in its impact), please make sure to first share discuss a design-doc with the team.

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

(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.

[lhb8125]

Please add label "expert review" when this PR is ready, or the reviewers won't get notifications.

[yanring]

Hi @yashaswikarnati, please include the main PR link in the desc.

[yashaswikarnati]

Hi @yashaswikarnati, please include the main PR link in the desc.

#2031 (this needs to go in first)

@yanring we cant yet create an MR to main as the one before this is not merged to main yet. Could we continue with the review and in the meanwhile we can create an MR to main

[shifangx]

Hi, @yashaswikarnati, could you please add some documentation to this PR?
For example, you could include some excerpts from the design document. https://docs.google.com/document/d/17FU3-CEAzob6lR40AY3LVCy57HHPsZZny8K4Z78l7ug/edit?tab=t.0#heading=h.bh7bns6bwmiq

[yashaswikarnati]

@yanring @lhb8125 gentle reminder on the review

[yanring]

@yanring @lhb8125 gentle reminder on the review

Thanks for the reminder. I'm actually waiting for a review from @lhb8125 and @shifangx to help take a look

[lhb8125]

Are we sure that the tensor must be 2D or 3D? Should we catch some exception here?

[lhb8125]

Can we move the logic of parsing input and output dict outside of the backward_step()?

[lhb8125]

@yashaswikarnati Could you add a brief description of the overall roadmap? I think this is a big feature that contains several MRs, so which one has been merged and which one is WIP? What's the role of the current PR in the whole feature?

[yanring]

Please ensure we have a passing functional test pipeline on GitLab before merging.