How can I get the "run_config.yaml" file when training with Megatron-LM?

[GHGmc2]

Hi,

I pre-trained a Qwen3-30B-A3B like model from scratch with Megatron-LM, and want to convert it to HuggingFace format with cmd:

$ python checkpoint_conversion.py export \
--hf-model Qwen/Qwen3-30B-A3B-Base \
--megatron-path /path/to/mcore_ckpt \
--hf-path /path/to/save

And I got error like:

  File "/megatron-bridge/examples/models/checkpoint_conversion.py", line 170, in export_megatron_to_hf
    raise FileNotFoundError(
FileNotFoundError: Could not find run_config.yaml in /path/to/mcore_ckpt. Please ensure this is a valid Megatron checkpoint.

After checking my ckpt files, I found that there's not a "run_config.yaml" file in them. How can I get the "run_config.yaml" file when training with Megatron-LM?

Thanks!