def test_updating_dataset_increases_its_size(self, dataset):
labels = types.TikTensorBatch(
[types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0))])
data = types.TikTensorBatch(
[types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0))])
dataset.update(data, labels)
assert 1 == len(dataset)
def test_access_by_index_to_deleted_element_is_allowed(self, dataset):
labels = types.TikTensorBatch(
[types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0))])
data = types.TikTensorBatch(
[types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0))])
dataset.update(data, labels)
dataset.remove((0, 0))
assert dataset[0]
def simple_dataset(self, dataset):
labels = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
data = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
dataset.update(data, labels)
return dataset
def test_removing_entries_from_dataset(self, dataset):
labels = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
data = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
dataset.update(data, labels)
assert 2 == len(dataset)
dataset.remove((0, 0))
assert 1 == len(dataset)
def test_access_by_index(self, dataset):
first_label = torch.Tensor(np.arange(9).reshape(3, 3))
first_data = torch.Tensor(np.arange(1, 10).reshape(3, 3))
labels = types.TikTensorBatch([
types.TikTensor(first_label, id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
data = types.TikTensorBatch([
types.TikTensor(first_data, id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
dataset.update(data, labels)
ret_data, ret_label = dataset[0]
assert torch.equal(first_label, ret_label)
assert torch.equal(first_data, ret_data)
def test_updating_removed_entries_recovers_them(self, dataset):
labels = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
data = types.TikTensorBatch([
types.TikTensor(np.ones(shape=(3, 3)), id_=(0, 0)),
types.TikTensor(np.ones(shape=(3, 3)), id_=(1, 0))
])
dataset.update(data, labels)
assert 2 == len(dataset)
dataset.remove((0, 0))
dataset.update(data, labels)
assert 2 == len(dataset)