Custom Translator full training job has been stuck in the “Training – Running” state for more than 48 hours

Hi TS

you're facing a frustrating situation with your Azure Custom Translator job being stuck in the “Training – Running” state for over 48 hours. Here are some steps you can take to address this issue:

Check Job Status: Sometimes training jobs can take longer than expected, especially with larger datasets. However, if it seems unusual, it could be stuck.

Cancel the Job: If possible, try to cancel the training job again from the Azure portal under your Custom Translator project. If the portal doesn’t let you cancel it, this may indicate a backend issue.

Using REST API: If cancellation via the portal fails, consider using the REST API to forcefully delete the job. The API provides a DELETE function that may allow you to remove the stuck job. You can find more information about managing models with the API here: Document Models - Delete Model.

Create a New Training Job: If the stuck job cannot be canceled, you could consider uploading your data again and starting a new training job under a different name.

Open a Support Case: If you continue to experience issues after trying the above steps, it's best to reach out to Azure support for further assistance.

I hope this helps resolve your issue! If you have any more details or specific questions, feel free to ask!

Reference Documentation

• Train a Custom Translator model
• Custom Translator frequently asked questions
• Document Models - Delete Model
• Quickstart: Build, publish, and translate with Custom Translator

Thank you!