feat: support override HF model name in convert_megatron_to_hf

[dhineshkumar-r]

…F format.

What does this PR do ?

Enables a way to override hf_model_name when converting checkpoints from megatron to HF format. This is useful for models like GPT-OSS whose base checkpoint precision(mxfp4) is different from supported export precision(bfloat16) in Megatron-Bridge, Ref.

Issues

List issues that this PR closes (syntax):

closes #2124

Usage

If openai/gpt-oss-20b is finetuned in bfloat16 precision and checkpoints are stored in megatron format, the override argument can be used to pass the supported unsloth/gpt-oss-20b-BF16 hf model name to use the config corresponding to bf16 precision.

uv run --extra mcore python examples/converters/convert_megatron_to_hf.py \
  --config <path_to_gpt_oss_20b_megatron_ckpt>/config.yaml \
  --hf-model-name unsloth/gpt-oss-20b-BF16 \
  --megatron-ckpt-path <path_to_gpt_oss_20b_megatron_ckpt>/policy/weights/iter_xxxxx \
  --hf-ckpt-path <path_to_save_hf_ckpt>

Before your PR is "Ready for review"

Pre checks:

• [Y] Make sure you read and followed Contributor guidelines
• [N] Did you write any new necessary tests?
• [N] Did you run the unit tests and functional tests locally? Visit our Testing Guide for how to run tests
• [Y] Did you add or update any necessary documentation? Visit our Document Development Guide for how to write, build and test the docs.

Additional Information

• Although there are not unit tests, verified the change works as expected locally by running above mentioned command with gpt-oss-20b checkpoint finetuned in bfloat16 precision. Did not notice any layers missing warnings.

• ...

[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.

[yuki-97]

thanks @dhineshkumar-r , changes lgtm.

could you help to add what you mentioned in the PR description to docs/design-docs/checkpointing.md, this will help people who meet the same situation.

[dhineshkumar-r]

Done. Please take a look.
cc: @yuki-97

[yuki-97]

/ok to test d162d13

[yuki-97]

hi @dhineshkumar-r , there's a lint check fail, could you use pre-commit run --all-files to fix it?

[dhineshkumar-r]

Yes, I don't see it fail anymore. Please let me know if anything else.

[yuki-97]

thanks, let me re-trigger CI.

[yuki-97]

/ok to test 3ea26b7