When I tried to fine tune pre-trained MobileNetV2 SSD FPN-Lite 320x320 object detection model available on the portal with my custom data of hundreds of images, after few epochs, the train stops with the error saying that training reached the maximum compute of free tier plan. that implies that I can use free tier plan only if I want to fine tune on few tens of images, not larger dataset.
Usually Free tier should be enough for casual usage, or study.
Review our increasing model performance guide, and share the project config / dataset size for review perhaps you have a setting thats causing this:
Best
Eoin