A machine learning engineer has deployed a model recommender using MLflow Model Serving. They now want to query the version of that model that is in the Production stage of the MLflow Model Registry.
Which of the following model URIs can be used to query the described model version?
To query the version of a model that is in the Production stage using MLflow Model Serving, the correct model URI should include the '/model/' prefix followed by the model name and the stage. Therefore, 'https://<databricks-instance>/model/recommender/Production/invocations' is the appropriate URI format.
The correct answer is A. https://<databricks-instance>/model-serving/recommender/Production/invocations. This URI points to the model named “recommender” that is in the “Production” stage of the MLflow Model Registry on the specified Databricks instance. The /invocations endpoint is used to make prediction requests to the model. Please note that <databricks-instance> should be replaced with the URL of your Databricks instance.
Correct answer is E. Because the URI should contain "/model/" and not "/model-serving/".
E is correct.
Doc: https://www.databricks.com/blog/2020/06/25/announcing-mlflow-model-serving-on-databricks.html