You can monitor all API calls to a deployed model. Navigate to the Deployment view and click the Monitoring tab.
The three graphs show:
API calls over time
All API calls during the specified time frame.
Predictions over time
Individual predictions served by the model over time.
Errors over time
All deployment API calls that respond with an error during the specified time frame.
This can be any kind of call that responds with an error. For example:
If the wrong type of input value is sent to the deployment.
There are issues with the model.
You’ve exceeded the number of predictions in your subscription plan.