Downloading Model Repositories for an air-gapped environment

To use Models from NVIDIA NGC and Hugging Face, the Administrator must download Model artifacts from these sources on specially networked hosts.

Downloading a HuggingFace model

  1. Download the Llama-3.1-Nemotron-70B-Instruct-HF Model from Hugging Face to your local file system with the following command:
    python3 import_to_airgap.py -do -rt hf -t hf_hVQbUCkpCicZYjnqsNAfafafafafafaAEkj -p $PWD/models -ri Nvidia/Llama-3.1-Nemotron-70B-Instruct-HF
    

    The download includes all Model files along with metadata in the specified destination directory.

  2. Download a different Hugging Face Model to your local file system with the following command:
    python3 import_to_arigap.py -do -rt hf -t <your-hf-token> -p $PWD/models -ri meta-llama/Llama-2-70b-chat-hf
                            
    • You can obtain the ri argument for Hugging Face as follows:
      1. Open up the Hugging Face page at: https://huggingface.co/.
      2. Search for the required model.

        The page of the model displays.

      3. Click the icon next to the name of the model, and copy the model ID, that is the ri argument to be used.
    The download includes all Model files along with metadata in the specified destination directory.

Downloading NGC model

  • Download the NVIDIA NGC Model to your local file system with the following command:
    python3 import_to_airgap.py -do -rt ngc -p $PWD/models -ri nim/meta/llama-3_1-70b-instruct:0.11.1+14957bf8-h100x4-fp8-throughput.1.2.18099809 -ns ngc_spec.yaml

    The download includes all Model files along with metadata in the specified destination directory.