NGC Registry Setup

You can create an NGC registry by setting up your NVIDIA Enterprise Account account. This enables HPE Machine Learning Inferencing Software to pull your models directly from the NVIDIA NGC catalog.

Before You Start #

Register for an NVIDIA Enterprise Account. Finalization of this can take 24-48 hours after submission.
- Environment details:
  - NVIDIA GPUS: Any
  - Use Case: inferencing
  - Project Size: Any
  - Infrastructure: cloud
  - Deployment: kubernetes
Review the NVIDIA NIM for LLMS guide.
Install the NGC CLI on your local machine.

How to Set Up an NGC Registry #

1. Generate a Personal Key #

Go to Setup > Personal Keys.
Select + Generate Personal Key.
Provide inputs for the following:
- Key Name: A name for your key.
- Expiration: Select an expiration date.
- Services Included:
  - NGC Catalog
  - Private Registry
Select Generate Personal Key. It will look similar to the following:
```
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
```
Copy the key and store it in a secure location.

2. Configure NGC CLI #

Run the following command to configure the NGC CLI:
```
ngc config set
```

Provide the following inputs:

Enter API key [no-apikey]. Choices: [<VALID_APIKEY>, 'no-apikey']: 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
Enter CLI output format type [ascii]. Choices: ['ascii', 'csv', 'json']:
Enter org [no-org]. Choices: ['<ORG-NAME>']: 
Enter team [no-team]. Choices: ['no-team']:
Enter ace [no-ace]. Choices: ['no-ace']:

Test #

Authenticate your local docker client with the NGC registry to validate the configuration was successful.

Use the following credentials:

Username: $oauthtoken    
Password: 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

Login Succeeded

You’re now ready to add an NGC registry to HPE Machine Learning Inferencing Software.

NGC CLI Commands #

For an in-depth list of NGC CLI commands, see the NGC CLI documentation.