Machine Learning Extractor Error " An Error Occurred While Retrieving Capabilities From The Server"

News Knowledge Base
1

An error occurred while retrieving capabilities from the server. Please check the logs for more details. Error message received when attempting to use the Skill in the Machine Learning Extractor activity in Studio.

Issue Description:
When attempting to configure the Machine Learning Extractor activity in Studio, An error occurred while retrieving capabilities from the server. Please check the log for more details.

image.png



Root Causes:

  • Incompatible skill version deployed
  • On-Prem installations of AI Center using incorrect certificate for On-Prem Orchestrator


Troubleshooting Steps:

  1. Validate that the ML Skill was deployed with a trained version of the package. For example: When deploying an ML Skill for the OOB DocumentUnderstanding model, if the minor version that was chosen is 0, try redeploying with a trained minor version (1 or greater).
  2. Validate that the API Key that is being used for the Machine Learning Extractor Activity is valid. See where to find the API Key. (not applicable for On-Prem air-gapped installations of AI Center.)
  3. Validate that the skill's endpoint can be reached after making the skill public.
    1. Navigate to the ML Skill
    2. If the ML Skill has not been made public, modify the deployment to make the ML Skill public. (Making the ML Skill public is only needed for troubleshooting purposes, this can be changed back after troubleshooting.) For how to make an ML Skill public, see Make ML Skill Public/Private.
    3. After confirming the ML Skill has been made public, copy the URL for the skill.
    4. On the machine where the issue is occurring, open a browser and paste the ML Skill's public endpoint in the navigation bar.
    5. The expected message in the browser is 405: Method is not allowed. <- This is expected (If something other than this is displayed, there could be an issue with reaching the ML Skill from the machine) Capture a screenshot to share the results with UiPath Product Support later if needed.


Note: The following steps are only applicable for On-Prem AiCenter installations. This is not applicable for UiPath Cloud AiCenter
  1. Validate that the version of the ML Skill deployed is compatible with the On-Prem version of AI Center that is being used.
    1. Click on ML Packages
    2. Click on the Package that is being used for the ML Skill
    3. After the page for the ML Package opens, take a screenshot of the version that is currently being used.
    4. Click Upload new version in the top right corner
    5. On this page, scroll to the bottom and look for the "Minimum AI Fabric version" listed.
    6. The version of AI Center that is currently in use is visible in the AI Center Navigation Menu.
    7. Now refer back to the screenshot that was previously taken of the ML Package version. If the version that is deployed is higher than the minimum required version of AI Fabric that is currently deployed, a new ML Package will need to be deployed that meets the minimum requirement for the AI Fabric version deployed.

  1. If Orchestrator and AI Center are deployed On-Prem, make sure the correct Orchestrator certificate is being used for AI Center. **Not applicable for UiPath Cloud deployments of AI Center.


Information to Share with Support: If the above suggestion does not resolve the issue, share the following with the Product Support Team,

  1. Replicate the issue and provide the logs generated for today's date from the following location %userprofile%\appdata\local\uipath\logs
  2. Provide some details regarding how you trained and deployed the ML skill that you have created. Feel free to include screenshots as well. How the ML Skill is deployed? When you deployed the skill, what minor version did you select?
  3. Provide screenshots that were captured during the Troubleshooting steps as described above.
  4. Share a screenshot of the project's dependencies or share the project files (Xamls, project.json, etc) in a zipped folder.
  5. If AI Center is On-Prem, also share a screenshot of the version of AI Center version that is deployed and also specify if the installation is Air-Gapped.