Deploy an out-of-the-box ML package

This page describes how to deploy an out-of-the-box Document Understanding ML Package. For demonstration purposes, we are going to use the Invoices ML Package as an example.

Create an Invoices ML package

For online installation, the Invoices model is already included in the Out of the box packages section. Go to ML Packages > Out of the box packages > UiPath Document Understanding > Invoices, and click Submit.

For offline installation, please upload the Invoices ML Package as a Zip file. To download the models, contact your Account Manager, CSM, or Support. You will receive a download link per package.

Choose Json input type and the corresponding Python language.

Click Create.

Deploy the Invoices ML package as an ML Skill

Go to the ML Skills tab from the left sidebar of AI Center and create a new ML Skill. This can take up to 30 minutes.

Congrats! You have successfully deployed the Invoices ML Model on AI Center.

Once the ML Skill is created, you can directly use the private ML Skill in the Studio workflow. Or, if you have installed the online version of AI Center standalone and you want to use the public ML Skill, please follow the below instruction to get the public ML Skill endpoint (optional).

Once the ML Skill is created, double-click the ML Skill and go to Modify current deployment.

Switch the toggle on to make the ML Skill public. You may need to wait for a few minutes and refresh the page.

Once the ML Skill is ready, copy the URL, which is the endpoint of the Invoices ML Package for later use.

Note:

If you want to import in an On Premises instance an out-of-the-box ML package that was previously exported from a Cloud instance, then you need to do the following:

• use this procedure to export the ML package
• open the metadata.json file for this ML package.
• identify the imagePath field and replace its value with the value found in the metadata page for the used ML package. Make sure to use the metadata value not the du bundle value.
• save all changes and start the package import to the On premises instance.

The ML packages with 2022.10 version can be installed only on an AI Center v2022.10.0, or higher.