NVIDIA provides scripts to convert the external Griffin checkpoints HuggingFace format to .nemo format. The .nemo checkpoint will be used for SFT, PEFT.
Run the NeMo container.
Convert the HuggingFace Gemma model to .nemo model:
python3 /opt/NeMo/scripts/checkpoint_converters/convert_griffin_hf_to_nemo.py \
--input_name_or_path /path/to/griffin/checkpoints/hf/2b \
--output_path /path/to/griffin-2b.nemo