test(checkpointing): seed CPU initialization in checkpoint load test

[cuichenx]

• I, the PR author, have personally reviewed every line of this PR.

What does this PR do?

Seed both model constructions in test_checkpoint_load_without_microbatch_calculator with torch.manual_seed(123).

The fixture seeds CUDA RNGs, but this test initializes its model weights on the CPU. Without an explicit CPU seed, data-parallel ranks can construct different reference weights. The legacy torch checkpoint saves rank 0's replica, so other ranks compare the restored weights against a different reference. Earlier tests can accidentally align CPU RNG states and mask the failure.

This is a two-line, test-only change. The restored model is still zeroed before loading, and the separate pre-load seed of 987 and all weight/output/RNG assertions remain unchanged. No production code or copyright changes.

⚠️ 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 @NVIDIA/mcore-oncall.

Issue tracking

For PRs from open-source community contributors:

• New features: a linked issue is required. Please open a feature request and reference it here before submitting the PR.
• Small updates (bug fixes, minor improvements): a linked issue is recommended and will accelerate the PR review process.

Linked issue: Related to #7792, which introduced this test.

Contribution process

Pre-checks

• I have added relevant unit tests
• I have added relevant functional tests
• If this PR adds or changes a GPU kernel (Triton, jit_fuser/torch.compile, CUDA extension, TE or external-library dispatch, or a scatter/index accumulation), I have added or updated its bit-exact determinism test and registered it in tests/unit_tests/determinism/kernels/manifest.py (guide)
• I have added proper typing to my code Typing guidelines
• I have added relevant documentation
• I have run the autoformatter.sh on my PR

Existing unit-test setup is corrected; no new test cases, GPU kernels, or public APIs are added.

• Independent full-diff review: no findings.
• Four-GPU isolated regression and all 16 parametrizations: passed on every rank.
• Ordering robustness: 140 test executions per rank across seven runs (normal, reverse, three seeded shuffles, and normal/shuffled mixed checkpoint-test sequences), all passed on four H100s. Stress runs deliberately set distinct CPU RNG states per rank after fixtures before each target test; recorded hashes confirm the states differ.
• Negative control on unpatched base 3914483fd: the same adversarial setup reproduces the original weight-equality assertion on ranks 1–3. Fixed head: d0d34b404. Experiment scripts are external to the PR.
• git diff --check, scoped Black and Ruff checks pass.
• The full formatting wrapper cannot run on the workstation because Git 2.25.1 is below its required 2.31 minimum.

Code review

Feel free to message or comment @NVIDIA/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.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[nemo-automation-bot]

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/38821466909