When the deployment reaches the running state, it is available to make predictions. See the full list of states here.
running
Once deployed, a secure URL will become available to make predictions. This URL will be persisted until the deployment is deleted.
If basic authentication is enabled, the username and password must be passed-in with each request in order to reach the endpoint.
Gradient supports both REST and gRPC Deployment endpoints. See this articlearrow-up-right for more information on the differences.
Incoming requests will be automatically load-balanced in a round-robin fashion if multiple instances are running.
Last updated 5 years ago