def test__get_loader_encdec(setup_args):
setup_args.model = 'shallow_encdec'
train_loader, val_loader = get_dataloaders(setup_args)
assert isinstance(train_loader, torch.utils.data.DataLoader)
assert isinstance(train_loader.dataset, SeqInSeqOut)
assert isinstance(val_loader, torch.utils.data.DataLoader)
assert isinstance(val_loader.dataset, SeqInSeqOut)
def test__get_loader_fnn(setup_args):
setup_args.model = 'shallow_fnn'
train_loader, val_loader = get_dataloaders(setup_args)
assert isinstance(train_loader, torch.utils.data.DataLoader)
assert isinstance(train_loader.dataset, FlatInFlatOut)
assert isinstance(val_loader, torch.utils.data.DataLoader)
assert isinstance(val_loader.dataset, FlatInFlatOut)
import torch
import torch.optim as optim
from motornn.utils.parser import get_parser_with_args
from motornn.utils.helpers import (get_file_names, get_dataloaders, get_model,
get_loss_function, Log)
from motornn.utils.runner import Runner
parser = get_parser_with_args()
args = parser.parse_args()
weight_path, log_path = get_file_names(args)
print(weight_path, log_path)
logger = Log(log_path, 'w')
train_loader, val_loader = get_dataloaders(args)
model = get_model(args)
criterion = get_loss_function(args)
optimizer = optim.SGD(model.parameters(), lr=args.lr)
runner = Runner(args.gpu, model, optimizer, criterion, train_loader,
val_loader)
best_smape = 1000
logger.write_model(model)
for epoch in range(args.epochs):
runner.set_epoch_metrics()
train_metrics = runner.train_model()