Manage Customization Configuration#
Use customization configurations to define the parameters for a customization job.
A customization configuration allows you to specify the target model you are customizing, along with training options, and their resources. This allows you to configure once, and re-use them for your Customization Jobs.
Administrators typically create customization configuration for your organization during the initial setup of the NeMo Microservices Platform. However, you can optionally create your own customization configurations via the API.
Tip
For Administrators: Many configurations may be created during initial setup but unavailable by default because their underlying customization targets are disabled. Users can discover configurations with disabled targets using the list configurations endpoint with filter[enabled]=false. These configurations can be activated by updating the corresponding target to set enabled=true using a PATCH API request. This allows for a controlled rollout of model capabilities.
Task Guides#
Perform common customization config tasks.
Tip
The value for CUSTOMIZER_BASE_URL will depend on the ingress in your particular cluster. If you are using the minikube demo installation, it will be http://nemo.test. Otherwise, you will need to consult with your own cluster administrator.
View all available customization configurations in your environment, filter by model or training type, and review details to select the right config for your fine-tuning jobs.
Retrieve details for a specific customization configuration, including all parameters and settings, using the API.
Create a new customization configuration by specifying the target model, training options, and resource requirements via a POST API call.
Update an existing customization configuration to change its description, training parameters, or other settings using a PATCH API request.