Closed kamangir closed 10 months ago
The notebooks are outdated; they use a previous version of the library, but when installing via pip, utils
causes issues. To use the notebook, you need to include the import
import segmentation_models_pytorch.utils
If you try to use the version of smp
installed with pip, then you'll find that things are different enough that the notebook won't easily work. For example, the losses are different and expect different input from the dataloader, and so are the metrics. If you use the notebook version, you cannot also use the most updated version of smp
(at least the metrics, you can still use the updated losses after changing the dataloader, but not both)
The reason being is that if you try to use the Trainer
class to train and validate per epoch, you have to adjust how you prepare the sample in the dataloader. I'm finding that you have to modify the samples in such a way that they are not compatible such that both the losses, and the metrics can be calculated in Trainer
. @qubvel am I mistaken?
This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 7 days.
This issue was closed because it has been stalled for 7 days with no activity.
error in https://github.com/qubvel/segmentation_models.pytorch/blob/master/examples/cars%20segmentation%20(camvid).ipynb