def test_one_step_tensor_delay_buffer():
"""Test a buffer which delays the data one step"""
shape = (3, 2, 1)
creator = AllocatingCreator(device='cpu')
buffer = create_delay_buffer(creator, True, shape)
no_delay = create_delay_buffer(creator, False, shape)
buffers = [buffer, no_delay]
nans = creator.full(shape, FLOAT_NAN)
a = creator.ones(shape)
b = creator.ones(shape) * 2
c = creator.ones(shape) * 3
d = creator.ones(shape) * 4
_push(buffers, a)
assert same(buffer.read(), nans)
assert same(no_delay.read(), a)
_push(buffers, b)
assert same(buffer.read(), a)
assert same(no_delay.read(), b)
_push(buffers, c)
assert same(buffer.read(), b)
assert same(no_delay.read(), c)
_push(buffers, d)
assert same(buffer.read(), c)
assert same(no_delay.read(), d)
def test_sample_learning_batch_combinations(self, method, flock_indices,
elements_written):
flock_size = 3
creator = AllocatingCreator('cpu')
f_size = len(flock_indices) if flock_indices is not None else 3
buffer = SPFlockBuffer(creator,
buffer_size=20,
n_cluster_centers=3,
flock_size=flock_size,
input_size=5)
if flock_indices is not None:
buffer.set_flock_indices(
creator.tensor(flock_indices, dtype=torch.int64))
buffer.total_data_written[:] = elements_written
buffer.clusters.stored_data[:, :elements_written] = creator.tensor(
[0, 1, 0])
buffer.inputs.stored_data[:, :elements_written, :] = creator.tensor(
[1.3, 0.2, 0.6, 0.4, 0.1])
# use some dummy value here to check that it is rewriting all the lines in res
dummy_value = -2.1
sampled_data = creator.full((f_size, elements_written, 5),
fill_value=dummy_value)
buffer.sample_learning_batch(elements_written, sampled_data, method)
assert (sampled_data == dummy_value).any().item() == 0
def _create_expected_images_tensor(creator: AllocatingCreator):
device = creator.device
dtype = get_float(device)
expected_images_tensor = creator.full([3, 2, 3, 3],
dtype=dtype,
device=device,
fill_value=1.0)
expected_images_tensor[0, 1, 2, :] = creator.tensor([1.0, 0.0, 0.0])
expected_images_tensor[1, 1, 2, :] = creator.tensor([0.0, 1.0, 0.0])
expected_images_tensor[2, 1, 2, :] = creator.tensor([0.0, 0.0, 1.0])
return creator.cat([expected_images_tensor, expected_images_tensor])