Model Serving - OpenShift Data Science
Design type: Details page | Product area: OpenShift AI
Author: Rachel Lombard
Last edit: September 28, 2023
Overview
Model serving involves taking a machine learning model and deploying it into a production environment. Once a model is deployed, it can accept input data and provide predictions based on that data. Users can monitor and detect issues through the model server table. Model serving can be seen in both the Project details page and in the left navigation item. Once a model has been deployed, the user can view all served models from all projects in the ‘Model serving' section of the left navigation item.
Adding a model server
Empty state
Users can add a model server in a project by clicking on the ‘Add model server’ button. An icon indicating an empty state and help text prompts the user to add a server to deploy a model.
An example of adding a model server empty state within Projects details page.
Modal
A model server modal will appear on screen for the user to enter model server information.
An example of add model server modal.
Deploying a model
Empty state
Once a server is added, the user can now deploy a model. The user needs to click on the ‘Deploy model’ button in order to do so.
An example of deploying a model empty state.
Modal
A modal will appear to add necessary information for model deployment.
An example of a modal when deploying a model.
Models and model servers
Details page
Once a model is deployed, users can access model information by switching between tabs of Deployed models and Model Server name.
An example of an expanded tab model server table.
An overview of models can be seen in the deployed models table where users can edit or delete deployed models.
An example of deployed models table viewed through the left nav item.