UiPath Orchestrator

The UiPath Orchestrator Guide

ML Packages

A machine learning package is a .zip file that contains all the code/metadata needed to run your model. Such an archive must include the model code, a requirements.txt file specifying Python dependencies and the functions needed to call the model. In order to be used within your workflows in Studio, you first have to deploy them as ML skills in your Orchestrator tenant.

It is recommended that users acting as Data Scientists handle ML packages. More details about recommended user personas here.

ML packages are somewhat similar to packages for your RPA workflows in Orchestrator. The ML Packages page enables you to view all the available versions of a package, along with their statuses, and change logs. Here you can upload new packages or new versions for existing ones, delete undeployed packages, or view available information about them.

Note:

You cannot perform external network requests from inside an ML Package.

Uploading ML Packages

Important!

Before uploading ML packages, make sure they are zipped in the right structure. Read about the correct .zip file structure here.

Follow these steps to upload an already created ML package:

  1. In the ML Packages page, click the Upload button. The Upload Package window is displayed.
  2. In the Upload Package window, click Browse to select the desired .zip file, or drag & drop the file anywhere within the window.
  3. Enter a name for your model.
  4. Select the input description type from the drop-down. The possible options are:
    • json
    • file
    • files
  5. (Optionally) Enter a description for the model's input.
  6. (Optionally) Enter a description for the model's output.
  7. Select the language from the drop-down. The possible options are:
    • Python 3.6
    • Python 3.7
    • Python 2.7
    • Mojo Pipeline
  8. (Optionally) Enter a description for the model.
  9. Select whether the machine learning model requires a GPU, by default it is set to No.
  10. Click Upload to upload the package or Cancel to abort the process. The Upload Package window is closed and the ML package is uploaded and displayed in the ML Packages page. It may take a few minutes before your upload is propagated.

Uploading New Model Versions

Follow these steps to upload a new version for an already uploaded ML package:

  1. In the ML Packages page, click the More Actions button next to a ML package and select Upload new model version. The Upload Package window is displayed, with most of the fields prefilled with the information you provided at the time when you first uploaded that package.
  2. In the Upload Package window, click Browse to select the desired .zip file, or drag & drop the file anywhere within the window.
  3. (Optionally) Update the existing information in the following fields:
    • Input description
    • Output description
    • Language.
      You can't change the model name and input description type, as they are specific to the already uploaded ML package.
  4. (Optionally) In the ChangeLog field, enter what has changed.
  5. Select whether the machine learning model requires a GPU, by default it is set to No.
  6. Click Upload to upload the new version for the existing uploaded package or Cancel to abort the process. The Upload Package window is closed and the new version of the ML package is uploaded. It may take a few minutes before your upload is propagated.

The new version of the package is not visible directly in the ML Packages page. You can view its information within the ML Package Versions window for that package.

Displaying Release Notes for one Package Version

  1. In the ML Packages page, click the More Actions button next to a ML package and select Info about the model. The ML Package Versions window is displayed for the selected package.
  2. On the Versions tab you can see all versions of the ML package displayed.
  1. For the desired version, click More details. The Release Notes window is displayed, enabling you to see the information that was added for the respective version.

Note that you have the option to remove all the undeployed versions of a package from the Versions tab.

Displaying Release Notes for all Package Versions

  1. In the ML Packages page, click the More Actions button next to a ML package and select Info about the model. The ML Package Versions window is displayed for the selected package.
  2. Click the Change Log tab. You can see all versions of the package, and for each, the corresponding release note containing information related to what has changed in-between versions.

Viewing the Arguments of a Package

  1. In the ML Packages page, click the More Actions button next to a ML package and select Info about the model. The ML Package Versions window is displayed for the selected package.
  1. Click Info about the version for the desired version. The input and output values of the selected package version are displayed. Please note that you cannot edit the values.

Removing All Undeployed Models

ML packages can only be deleted from Orchestrator if they are not deployed within an ML skill.

  1. In the ML Packages page, click the More Actions button next to a ML package and select Remove All Undeployed. A confirmation window is displayed.
  2. In the confirmation window, click Yes to delete all undeployed versions of the selected package. If a package version is part of an ML skill (it is active), it is NOT going to be deleted. If all the versions are inactive, they are all deleted.

OR

  1. In the ML Packages page, click the More Actions button next to a ML package and select Info about the model. The ML Package Versions window is displayed for the selected package.
  2. On the Versions tab, click Remove All Undeployed. A confirmation window is displayed.
  1. In the confirmation window, click Yes to delete all undeployed versions of the selected package. If a package version is part of an ML skill (it is active), it is NOT going to be deleted. If all the versions are inactive, they are all deleted.

Updated 2 months ago



ML Packages


Suggested Edits are limited on API Reference Pages

You can only suggest edits to Markdown body content, but not to the API spec.