I'd like to train a Yolov5 model as an Azure ML job (Python SDK v2). Ideally, I'd like to track all metrics during training with MLFlow, so that I could compare the runs in the native Azure ML workspace.
What's the best way to do this? I only know how to start training in the terminal by running yolov5 train --data ..., so I'm not sure how to set callbacks. Is it maybe possible to train yolo within Python using torch (if so, how?) and configure callbacks there?
My last resort would be to track the performance only after model training.
Do you have any advice?
Thanks!
