this is a known service-side issue that sometimes happens with custom vision. the training process gets stuck in a loop or hangs on a backend error. since you cannot cancel or reset it, the only real fix is to contact azure support directly.
you need to open a support ticket for your custom vision project. be sure to give them your project id, project name, and the region where your custom vision resource is located. they have tools on their end to force-stop the training and get your project back to a usable state.
while you wait for them, its worth trying to create a brand new project and resource in a different azure region. sometimes a fresh start in a new location can bypass whatever glitch is happening. just use a small subset of your data to test if the new project trains correctly.
also, check the service health status in your azure portal. there might be an ongoing incident affecting custom vision that is causing these delays.
really hope they can get this resolved for you quickly. waiting that long is completely unacceptable.
Best regards,
Alex
and "yes" if you would follow me at Q&A - personaly thx.
P.S. If my answer help to you, please Accept my answer