Configuration
A deployment configuration defines how a model is served after training.
Create a configuration before deploying a model. The deployed model then uses the configuration to determine the serving mode and resources.
Choose a deployment type
Deployment configurations are created as either SERVERLESS or REALTIME.
Serverless configurations use memory size and maximum concurrency. Real-time configurations use a serving machine type and machine count.
Create a serverless configuration
Create a serverless configuration when the model only needs to serve requests on demand.
Save the returned configuration id. You will use it when deploying a model.
Create a real-time configuration
Create a real-time configuration when the model needs to stay ready for low-latency requests.
Before creating the configuration, retrieve the available serving machine types.
Then create the deployment configuration with deployment_type set to REALTIME.
Save the returned configuration id. You will use it when deploying a model.
Retrieve configurations
Retrieve deployment configurations when you need to reuse an existing configuration.
To inspect a specific configuration, retrieve it by id.
Update a configuration
Update a deployment configuration when you need to rename it.
Delete a configuration
Delete a deployment configuration when it should no longer be used by future deployments.

