def test_round():
for lib, call in helpers.calls:
assert np.array_equal(call(ivy_gen.round, ivy_gen.array([0.3], f=lib)),
np.round(np.array([0.3])))
assert np.array_equal(
call(ivy_gen.round, ivy_gen.array([[0.51]], f=lib)),
np.array([[1.]]))
def test_linear():
fname = os.path.join(this_file_dir, 'runtime_analysis/{}/layers/linear.txt'.format(DIM))
if os.path.exists(fname):
os.remove(fname)
for lib, call in [(l, c) for l, c in helpers.calls if c not in [helpers.tf_graph_call, helpers.mx_graph_call]]:
time_lib = LIB_DICT[lib]
append_to_file(fname, '{}'.format(lib))
x0 = ivy_gen.array([[random.uniform(0, 1) for _ in range(DIM)]], f=lib)
weight = ivy_gen.array([[random.uniform(0, 1) for _ in range(DIM)],
[random.uniform(0, 1) for _ in range(DIM)]], f=lib)
bias = ivy_gen.array([random.uniform(0, 1), random.uniform(0, 1)], f=lib)
ivy_layers.linear(x0, weight, bias, num_hidden=2, f=lib)
ivy_layers_w_time.linear(x0, weight, bias, num_hidden=2, f=time_lib)
TIMES_DICT.clear()
for _ in range(100):
log_time(fname, 'tb0')
ivy_layers_w_time.linear(x0, weight, bias, num_hidden=2, f=time_lib)
log_time(fname, 'tb4', time_at_start=True)
log_time(fname, 'tt0')
ivy_layers.linear(x0, weight, bias, num_hidden=2, f=lib)
log_time(fname, 'tt1', time_at_start=True)
write_times()
append_to_file(fname, 'end of analysis')
def test_floor():
for lib, call in helpers.calls:
assert np.array_equal(call(ivy_gen.floor, ivy_gen.array([0.3], f=lib)),
np.floor(np.array([0.3])))
assert np.array_equal(
call(ivy_gen.floor, ivy_gen.array([[0.7]], f=lib)),
np.floor(np.array([[0.7]])))
def test_ceil():
for lib, call in helpers.calls:
assert np.array_equal(call(ivy_gen.ceil, ivy_gen.array([0.3], f=lib)),
np.ceil(np.array([0.3])))
assert np.array_equal(
call(ivy_gen.ceil, ivy_gen.array([[0.7]], f=lib)),
np.ceil(np.array([[0.7]])))
def test_unstack():
for lib, call in helpers.calls:
if call is helpers.mx_graph_call:
# mxsymbolic split returns either list or tensor depending on number of splits
continue
x = np.swapaxes(np.array([[0.]]), 0, 0)
true = [np.array(item) for item in x.tolist()]
pred = call(ivy_gen.unstack,
ivy_gen.array([[0.]], f=lib),
0,
num_outputs=1)
assert _reduce(
_mul,
[np.array_equal(pred_, true_)
for pred_, true_ in zip(pred, true)], 1) == 1
x = np.swapaxes(np.array([[[0.]]]), 0, 0)
true = [np.array(item) for item in x.tolist()]
pred = call(ivy_gen.unstack,
ivy_gen.array([[[0.]]], f=lib),
0,
num_outputs=1)
assert _reduce(
_mul,
[np.array_equal(pred_, true_)
for pred_, true_ in zip(pred, true)], 1) == 1
def test_tile():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_gen.tile, ivy_gen.array([[0.]], f=lib), [1, 2]),
np.tile(np.array([[0.]]), [1, 2]))
assert np.array_equal(
call(ivy_gen.tile, ivy_gen.array([[[0.]]], f=lib), [1, 2, 3]),
np.tile(np.array([[[0.]]]), [1, 2, 3]))
def test_split():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_gen.split, ivy_gen.array([[0., 1.]], f=lib), 2, -1),
np.split(np.array([[0., 1.]]), 2, -1))
assert np.array_equal(
call(ivy_gen.split, ivy_gen.array([[[0., 1.]]], f=lib), 2, -1),
np.split(np.array([[[0., 1.]]]), 2, -1))
def test_cast():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_gen.cast, ivy_gen.array([0], f=lib), 'float32'),
np.array([0]).astype(np.float32))
assert np.array_equal(
call(ivy_gen.cast, ivy_gen.array([[0]], f=lib), 'float32'),
np.array([[0]]).astype(np.float32))
def test_abs():
for lib, call in helpers.calls:
assert np.allclose(call(ivy_gen.abs, ivy_gen.array([-0.3], f=lib)),
np.array([0.3]),
atol=1e-6)
assert np.allclose(call(ivy_gen.abs, ivy_gen.array([[-0.7]], f=lib)),
np.array([[0.7]]),
atol=1e-6)
def test_logical_not():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_logic.logical_not, ivy_gen.array([True, True], f=lib)),
np.logical_not(np.array([True, True])))
assert np.array_equal(
call(ivy_logic.logical_not, ivy_gen.array([[0.]], f=lib)),
np.logical_not(np.array([[0.]])))
def test_clip():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_gen.clip, ivy_gen.array([0.], f=lib), 0, 1),
np.clip(np.array([0.]), 0, 1))
assert np.array_equal(
call(ivy_gen.clip, ivy_gen.array([[0.]], f=lib), 0, 1),
np.clip(np.array([[0.]]), 0, 1))
def test_reduce_mean():
for lib, call in helpers.calls:
assert np.array_equal(call(ivy_red.reduce_mean, ivy_gen.array([1., 2., 3.], f=lib), 0, True),
np.mean(np.array([1., 2., 3.]), keepdims=True))
assert np.array_equal(call(ivy_red.reduce_mean, ivy_gen.array([1., 2., 3.], f=lib), (0,), True),
np.mean(np.array([1., 2., 3.]), keepdims=True))
assert np.array_equal(call(ivy_red.reduce_mean, ivy_gen.array([[1., 2., 3.]], f=lib), (0, 1), True),
np.mean(np.array([[1., 2., 3.]]), keepdims=True))
def test_pinv():
for lib, call in helpers.calls:
if call is helpers.mx_graph_call:
# mxnet symbolic does not support pinv
continue
assert np.allclose(call(ivy_linalg.pinv, ivy_gen.array([[1., 0.], [0., 1.], [1., 0.]], f=lib)),
np.linalg.pinv(np.array([[1., 0.], [0., 1.], [1., 0.]])), atol=1e-6)
assert np.allclose(call(ivy_linalg.pinv, ivy_gen.array([[[1., 0.], [0., 1.], [1., 0.]]], f=lib)),
np.linalg.pinv(np.array([[[1., 0.], [0., 1.], [1., 0.]]])), atol=1e-6)
def test_random_shuffle():
for lib, call in helpers.calls:
call(ivy_rand.seed, 0, f=lib)
first_shuffle = call(ivy_rand.shuffle, ivy_gen.array([1, 2, 3], f=lib),
lib)
call(ivy_rand.seed, 0, f=lib)
second_shuffle = call(ivy_rand.shuffle, ivy_gen.array([1, 2, 3],
f=lib), lib)
assert np.array(first_shuffle == second_shuffle).all()
def test_concatenate():
for lib, call in helpers.calls:
assert np.array_equal(
call(ivy_gen.concatenate,
(ivy_gen.array([0.], f=lib), ivy_gen.array([0.], f=lib)), 0),
np.concatenate((np.array([0.]), np.array([0.])), 0))
assert np.array_equal(
call(ivy_gen.concatenate,
(ivy_gen.array([[0.]], f=lib), ivy_gen.array([[0.]], f=lib)),
0), np.concatenate((np.array([[0.]]), np.array([[0.]])), 0))
def test_argmin():
for lib, call in helpers.calls:
assert np.allclose(call(ivy_gen.argmin,
ivy_gen.array([-0.3, 0.1], f=lib)),
np.array([0]),
atol=1e-6)
assert np.allclose(call(
ivy_gen.argmin, ivy_gen.array([[1.3, -0.7], [0.1, 2.5]], f=lib)),
np.array([1, 0]),
atol=1e-6)
def test_norm():
for lib, call in helpers.calls:
assert np.array_equal(call(ivy_linalg.norm, ivy_gen.array([[1., 0.], [0., 1.]], f=lib), 1),
np.linalg.norm(np.array([[1., 0.], [0., 1.]]), 1, -1))
assert np.array_equal(call(ivy_linalg.norm, ivy_gen.array([[1., 0.], [0., 1.]], f=lib), 1, 1),
np.linalg.norm(np.array([[1., 0.], [0., 1.]]), 1, 1))
assert np.array_equal(call(ivy_linalg.norm, ivy_gen.array([[1., 0.], [0., 1.]], f=lib), 1, 1, True),
np.linalg.norm(np.array([[1., 0.], [0., 1.]]), 1, 1, True))
assert np.array_equal(call(ivy_linalg.norm, ivy_gen.array([[[1., 0.], [0., 1.]]], f=lib), 2),
np.linalg.norm(np.array([[[1., 0.], [0., 1.]]]), 2, -1))