def test__loader(setup_args):
full_load, train_samples, val_samples, metadata = load_data(setup_args)
x, y = loader(full_load,
train_samples[0],
metadata,
setup_args,
type='seq')
assert isinstance(x, np.ndarray)
assert isinstance(y, np.ndarray)
assert x.shape == (4, setup_args.window)
assert y.shape == (2, setup_args.window)
assert x.min() >= -1
assert x.max() <= 1
assert y.min() >= -1
assert y.max() <= 1
x, y = loader(full_load,
train_samples[0],
metadata,
setup_args,
type='flat')
assert y.shape == (2, )
def test__init(self, setup_args):
dataset, train_samples, _, metadata = load_data(setup_args)
dataloader = SeqInSeqOut(dataset, train_samples, metadata, setup_args)
assert isinstance(dataloader.samples, list)
assert isinstance(dataloader.full_load, dict)
assert isinstance(dataloader.metadata, dict)
def test__getitem__(self, setup_args):
dataset, train_samples, _, metadata = load_data(setup_args)
dataloader = SeqInSeqOut(dataset, train_samples, metadata, setup_args)
inp_seq, out_seq = dataloader.__getitem__(0)
assert isinstance(inp_seq, np.ndarray)
assert isinstance(out_seq, np.ndarray)
assert len(inp_seq.shape) == 2
assert inp_seq.shape == (4, 100)
assert len(out_seq.shape) == 2
assert out_seq.shape == (2, 100)
def get_dataloaders(args):
dataset, train_samples, val_samples, metadata = load_data(args)
preloader_class = _get_prelaoder_class(args)
print('Train Samples ', len(train_samples))
print('Val Samples ', len(val_samples))
train_preloader = preloader_class(dataset, train_samples, metadata, args)
train_loader = DataLoader(train_preloader,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers)
val_preloader = preloader_class(dataset, val_samples, metadata, args)
val_loader = DataLoader(val_preloader,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers)
return train_loader, val_loader
def test__load_data(setup_args):
full_load, train_samples, val_samples, metadata = load_data(setup_args)
assert full_load
assert isinstance(full_load, dict)
for k in full_load.keys():
assert 'train' in k or 'val' in k
assert '0001.pkl' in k
first_data = full_load[list(full_load.keys())[0]]
assert isinstance(first_data, dict)
assert 'voltage_d' in first_data
assert 'voltage_q' in first_data
assert 'current_d' in first_data
assert 'current_q' in first_data
assert 'statorPuls' in first_data
assert 'speed' in first_data
assert 'torque' in first_data
assert 'time' in first_data
assert 'reference_torque_interp' in first_data
assert 'reference_speed_interp' in first_data
assert 'reference_torque' in first_data
assert 'reference_speed' in first_data
assert 'torque_time' in first_data
assert 'speed_time' in first_data
assert isinstance(train_samples, list)
assert isinstance(val_samples, list)
assert train_samples
assert val_samples
assert isinstance(train_samples[0], list)
assert isinstance(val_samples[0], list)
assert len(train_samples[0]) == 4
assert isinstance(metadata, dict)