Issue with Training Loop and Active Learning in MONAILabel+CVAT

[bilalcodehub]

Dear Team,

I’m using MONAILabel with CVAT for an active learning workflow. While the labels were successfully downloaded from CVAT, the training loop terminated because Docker couldn’t access the GPU. MONAILabel marked the CVAT task as DONE_<task_name>, but the training process didn’t complete, and no new task is being created in the active learning loop.

I restarted the MONAILabel Docker container (now with GPU access) and relaunched the server, but the training loop hasn’t resumed, nor has it created a new task.

Could you help me understand why the active learning loop isn’t resuming or generating new tasks after the training failure? What steps can I take to resolve this and ensure the loop continues seamlessly?

Thank you for your support!

Kind regards,
Bilal