Deploying a model using the user interface
Describes how to deploy a model using IBM Watson Machine Learning.
About this task
Once you're satisfied with its results, reliability, and performance, you can deploy a model inside Watson Studio using IBM Watson Machine Learning.
- In the model you want to deploy, open either Scenario 1 or the specific scenario you want to deploy.
Click the menu icon beside the scenario name and select Save as model for deployment.
If you have set any run configuration parameters, these will be used in deployment.
Specify a name for your model and add a description if required, then click
The model is available in the Models section of your project.
In your project, click the
Git actions icon to commit and push your changes to the master
You'll use the tag you associate with this commit to locate the model for deployment.
Switch to the Watson Machine Learning view using the top menu.
Create a new Project release.
Specify a name and a route, then select your project and the tag you used when you committed the model for deployment. Click Create.
- Open your new release and select the Assets tab, then select your deployable model and click add service to add a web service to it.
On the Create deployment page, specify a name for your deployment.
Allocate a number of cores and a memory size for the service, and use the load balancer to specify how many instances of your service you want to have available. Click Create to create the deployment.
- On the Deployments tab for your release, click the Launch icon.
- Select Enable from the three dots icon for
your deployment. Your model is ready to receive requests.
- You can access the token for your model by clicking your deployment
in the Project release page. A window opens where you can click your deployment token to copy it.
You can access information about your service on the Active environments tab.
Once you've deployed a model, refer to Submitting a job to a deployed model for details on how to execute it.