airpack_scripts.pytorch.run_training¶
Module Contents¶
-
airpack_scripts.pytorch.run_training.DEVICE¶
-
airpack_scripts.pytorch.run_training.train(network, dataloader, optimizer, criterion, scalar)¶ Train pytorch neural network and return loss and accuracy
- Parameters
network (torch.nn.Module) – torch model
dataloader (torch.utils.data.dataloader) – torch data loader
optimizer (torch.optim) – torch optimizer function
criterion (torch.nn) – torch loss function
scalar (int) – scalar integer to normalize data
- Returns
(train_loss, train_acc)- Return type
tuple
-
airpack_scripts.pytorch.run_training.validate(network, dataloader, criterion, scalar)¶ Train pytorch neural network and return loss and accuracy
- Parameters
network (torch.nn.Module) – torch model
dataloader (torch.utils.data.dataloader) – torch data loader
criterion (torch.nn) – torch loss function
scalar (int) –
- Param
scalar int to normalize data
- Returns
(val_loss, val_acc)- Return type
tuple
-
airpack_scripts.pytorch.run_training.main(data_folder, n_epoch=10)¶ Script used to train
airpack.pytorch.model.Network.Note
You may modify the parameters in this script to tune the hyperparameters of the
airpack.pytorch.model.Network().- Parameters
data_folder (Union[str, os.PathLike]) – Location of training data
n_epoch (int) – Number of epochs in training process
- Returns
Training accuracy
- Return type
float