Need help with training a YOLO model on a bounding boxes dataset in Pytorch

I have a very high quality dataset which I am using to train a YOLO model. I have only been able to get an accuracy of max 28% so far (transformations have already been applied), its a very simple model but needs to be adjusted so higher accuracy may be achieved quickly.

I have the model loaded in my notebook and looking for someone with experience in YOLO training to help me better the accuracy.