Publish

You can use the Publish section to:

• Create a project version: freeze the current models state into a project version.
• Deploy project version: make the models accessible from workflows.
• Automate your process: consume the product version in an automation and open Studio Desktop or integrate using APIs.
• Export dataset so you can use it in a different tenant or organization, or just save it as a backup.

Project version

A project version is a snapshot of the current models state. Since the Classifier and Extractor models are trained continuously as you upload data or interact with the model, you can use the Project version feature to freeze the current state of the models if you are satisfied with the performance. Once a project version is created, you can start using your models in a workflow.

Creating a new project version

If you are satisfied with the current model performance, or want to test the model behavior in a workflow, create a new project version.

1. Go to the Publish section in your project.
2. Click Create a project version.
3. Fill in the name of your project version in the Name field.
4. [Optional] Add a description for your project version.
5. Choose the models you want to be included from the Models to include drop-down list.
   Tip: Classification is automatically enabled when multiple document types are selected.
6. Choose the deployment type.
7. Click Create.

Once a project version is created, model training is automatically kicked off. Both Classifier and Extractor selected during process version creation are trained on all uploaded data. Depending on the model and the size of the dataset, this can take from a few minutes to several hours.

Tip: You can check the model version used in your project version. To do so, select the three-dot icon ⋮ next to the project version and choose Edit version. You can check the model version from the Selected models section.

Deploying a project version

After a project is created, the project version can have one of the following status:

• Training: Models in the project are being trained. This is a transitory state.
• Trained: Models in the project are trained.
• Undeployed: Models trainining is complete, but the models are in idle state. Models are not consuming any resources and the project version cannot be referenced from, or used in workflows. Undeployed project versions are not available in Studio activities.
• Deployed: Models training is complete, and the are running. Models are consuming hardware resources and the project version can be referenced from and used in workflows.

To deploy a project version, click on the toggle from the Deployed column.

Automate your process

Once a project version is deployed, you can use the models in workflows.

Note: If you have 500 projects created on your tenant and use one of the following activities, any projects after 500 will not be visible in UiPath Studio or Studio Web and cannot be used:

• Extract Document Data and Classify Document activities from the DocumentUnderstanding.Activities package.
• Document Understanding Project Classifier and Document Understanding Project Extractor from the IntelligentOCR.Activities package.

If you are using Document Understanding^TM Cloud APIs, you can also use the project in your API automation. You can do this easily be selecting Integrate via API.

Note: The Integrate via API button is only active if you have extraction models in your project version.

If you select Integrate via API, a new window with the code is displayed. Here, you have the option to select your preferred programming language and paste this code into your preferred editor. However, please remember to modify the specified variables before consuming the deployed project via REST APIs.

Export

You can export annotated datasets from your modern project and:

• Import them into another modern project from a different tenant or organization.
• Import them into a classic project. For more information on how to import datasets into a classic project, check the Document Understanding User Guide for classic projects.
  Note: Datasets exported from modern projects are compatible with classic projects only when training on model version 23.4.0 or newer.

Important: Only annotated and confirmed (green shield icon) are exported.

For more information on how to export datasets and migrate your modern projects, check the Migrating modern projects page.