[DAISY-gh]

What does this PR do ?

Fix fsdp_dtensor conversion for pretrained-only checkpoints

⚠️ 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.

Fix two bugs in convert-torch-dist-to-fsdp-dtensor that cause crashes when converting pretrained-only checkpoints (no optimizer state, e.g. HuggingFace-converted weights) to fsdp_dtensor format.

Changes

1. tools/checkpoint/checkpoint_inspector.py — AttributeError: 'Metadata' object has no attribute 'mcore_data'

convert_checkpoint() directly accesses metadata.mcore_data without checking if the attribute exists. mcore_data is a Megatron-Core extension that only appears in checkpoints containing distributed optimizer state. Pretrained-only checkpoints (e.g. converted from HuggingFace) have no optimizer state, so their metadata lacks this attribute.

Fix: Add hasattr(metadata, "mcore_data") guards before all unprotected accesses. This is consistent with the existing pattern already used elsewhere in the same file (lines 791, 804).

2. megatron/core/distributed/fsdp/src/megatron_fsdp/uneven_dtensor.py — RuntimeError: spec.tensor_meta.shape must be a tuple

split_dtensor() builds out_shape as a list (to allow element mutation at line 475), then passes it directly to DTensor.from_local(shape=out_shape). Newer PyTorch versions (e.g. PyTorch 2.6 in NeMo 26.02 container) enforce that shape must be a tuple.

Fix: Wrap with tuple(out_shape) before passing to DTensor.from_local().

Reproduction

Convert a HuggingFace-originated mcore torch_dist checkpoint (no optimizer state) to fsdp_dtensor

torchrun --nproc_per_node=4 --nnodes=1
tools/checkpoint/checkpoint_inspector.py
convert-torch-dist-to-fsdp-dtensor --swiglu
/path/to/mcore-torch-dist-ckpt/iter_0000000
/path/to/output-fsdp-dtensor
--param-to-param-group-map-json /path/to/empty-map.json

Contribution process

Pre-checks

• 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

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!

All PRs start as draft. If you open a non-draft PR, it will be automatically converted to draft.

Step 1: Mark PR as "Ready for Review"

1. When your PR is ready, click Ready for Review.
2. An oncall reviewer is auto-assigned and expert reviewers are notified based on your changes.
   • Some PRs may jump straight to step 2. This is determined by .github/CODEOWNERS.

⚠️ Only mark as ready once merge-conflicts are resolved and the CI is passing.
Final Review might get declined if these requirements are not fulfilled.

Step 2: Final Review

For PRs that change megatron/core, once all expert reviewers have approved, the Final Review label is applied automatically and final reviewers are assigned.

For PRs outside megatron/core, this step is skipped.

Step 3: Approved

Once all required reviewers have approved, the Approved label is applied automatically.

Merge

Any member of mcore-engineers will be able to merge your PR.

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.

[copy-pr-bot]

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]

This PR has been automatically converted to draft because all PRs must start as drafts.

When you are ready for review, click Ready for Review to begin the review process. This will:

1. Add the oncall reviewer (optional reviewer)
2. Add required review teams based on your changes

See the contribution guide for more details.

[shjwudp]

LGTM, thanks!

[shjwudp]

/ok to test d8c4cc9

[cspades]

Approving to unblock.

[cspades]

/ok to test 0af5a7d

[svcnvidia-nemo-ci]

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/23862729458