search

kuangliu / pytorch-retinanet

RetinaNet in PyTorch
992 stars 250 forks source link

loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.697 | avg_loss: 1.697 #59

Open Imagery007 opened 5 years ago

Imagery007 commented 5 years ago

$ python train.py --lr 0.001 ==> Preparing data..

Epoch: 0 /home/zhangchaohua/envs/pytorch/lib/python2.7/site-packages/torch/nn/functional.py:2351: UserWarning: nn.functional.upsample is deprecated. Use nn.functional.interpolate instead. warnings.warn("nn.functional.upsample is deprecated. Use nn.functional.interpolate instead.") /home/zhangchaohua/envs/pytorch/lib/python2.7/site-packages/torch/nn/functional.py:2423: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details. "See the documentation of nn.Upsample for details.".format(mode)) /home/zhangchaohua/envs/pytorch/lib/python2.7/site-packages/torch/nn/_reduction.py:49: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead. warnings.warn(warning.format(ret)) loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.697 | avg_loss: 1.697 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.391 | avg_loss: 1.544 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.472 | avg_loss: 1.520 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.385 | avg_loss: 1.486 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.475 | avg_loss: 1.484 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.300 | avg_loss: 1.453 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.414 | avg_loss: 1.448 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.418 | avg_loss: 1.444 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.468 | avg_loss: 1.447 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.393 | avg_loss: 1.441 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.466 | avg_loss: 1.444 loc_loss: 0.000 | cls_loss: 1.000 | train_loss: 1.588 | avg_loss: 1.456

Imagery007 commented 5 years ago

I just trained, why is my loss like this? I am using Pytorch-1.0.1.

Imagery007 commented 5 years ago

batch-size=8，After training for an hour： loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.485 | avg_loss: 0.577 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.497 | avg_loss: 0.577 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.471 | avg_loss: 0.577 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.451 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.547 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.447 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.560 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.496 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.498 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.366 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.594 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.652 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.628 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.529 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.428 | avg_loss: 0.576 loc_loss: 0.000 | cls_loss: 0.000 | train_loss: 0.447 | avg_loss: 0.576 What happened？

Imagery007 commented 5 years ago

I solved the problem with his loss.py. It looks normal now。 https://github.com/kuangliu/pytorch-retinanet/issues/52#issuecomment-489348555