adnormal loss for training process

roytseng-tw / Detectron.pytorch

A pytorch implementation of Detectron. Both training from scratch and inferring directly from pretrained Detectron weights are available.

MIT License

2.82k stars 567 forks source link

adnormal loss for training process #177

Open At1a8 opened 5 years ago

At1a8 commented 5 years ago

Hi, I trained my dataset (only one class) on your code and train the network following "e2e_faster_rcnn ResNet-50.yaml". However, I find that loss_cls is larger than loss_rpn_cls, which is not a ideal training process. Could you please give some advises, thank you for your time