Azure AI - Language Studio – Data Labeling – Auto-label with GPT option - Failed job - Missing error details

Question

I want to create a capability to identify some custom business text value from files using Azure AI.

So I created a project in Azure AI - Language Studio and uploaded my files into an Azure Storage Account Container.

The wizard started with data labeling (I have no label data file).

I tried to use auto-labelling by following this tutorial:

https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/how-to/use-autolabeling?tabs=autolabel-gpt

I selected “Auto-label with GPT option”.

I went through all the wizard steps and click Start job:

The job started but after some time it failed:

with this error:

One or more errors occurred. Corresponding activity ID: 65b4b7ac-9245-44b0-a1de-da11d63637c8.

No further details were provided.

Where could I find error details about this activity ID (65b4b7ac-9245-44b0-a1de-da11d63637c8) to be able to fix this error?

Answer 1

To troubleshoot this auto-labeling failure in Azure AI Language Studio, here are a few steps you can take to locate more details on the error:

Check Azure Activity Logs: Go to the Azure portal and navigate to Activity Logs under Monitor. Use the Activity ID (65b4b7ac-9245-44b0-a1de-da11d63637c8) to filter logs, which might provide more context about the error.

Inspect Logs in Language Studio (if available): Sometimes, additional logs or details might be available directly in Language Studio under the Jobs or Logs section related to data labeling or training jobs.

Look at the Storage Account Logs: If you’re using Azure Storage to store your input files, you can check the Diagnostics and Logs for the storage account to ensure files were accessed correctly.

Enable Diagnostic Settings for Detailed Logs: Enable Diagnostic settings for the Azure AI service associated with Language Studio. Set up logs to go to Log Analytics, Event Hub, or Storage Account for deeper insights.

Contact Azure Support: If you still cannot locate useful error details, consider reaching out to Azure Support with the Activity ID, as they can access backend logs that are not exposed to customers.

This will help pinpoint issues like missing configuration, permission issues, or potential misconfigurations in the auto-labeling setup.