def test_qn(loss, dtype, penalty, l1_strength, l2_strength, fit_intercept):
tol = 1e-6
X = np.array(precomputed_X, dtype=dtype)
if loss == 'sigmoid':
y = np.array(precomputed_y_log, dtype=dtype)
else:
y = np.array(precomputed_y_multi, dtype=dtype)
qn = cuQN(loss=loss, fit_intercept=fit_intercept, l1_strength=l1_strength,
l2_strength=l2_strength, tol=1e-8)
qn.fit(X, y)
print(qn.objective)
print(qn.coef_.copy_to_host())
if loss == 'sigmoid':
if penalty == 'none' and l1_strength == 0.0 and l2_strength == 0.0:
if fit_intercept:
assert (qn.objective - 0.40263831615448) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.1088872],
[2.4812558],
[0.7960136]]),
decimal=3)
else:
assert (qn.objective - 0.4317452311515808) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.120777],
[3.056865]]),
decimal=3)
elif penalty == 'l1' and l2_strength == 0.0:
if fit_intercept:
if l1_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.1088872],
[2.4812558],
[0.7960136]]),
decimal=3)
else:
assert (qn.objective - 0.44295936822891235) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.6899368],
[1.9021575],
[0.8057671]]),
decimal=3)
else:
if l1_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.120777],
[3.056865]]),
decimal=3)
else:
assert (qn.objective - 0.4769895672798157) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.6214856],
[2.3650239]]),
decimal=3)
# assert False
elif penalty == 'l2' and l1_strength == 0.0:
if fit_intercept:
if l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.1088872],
[2.4812558],
[0.7960136]]),
decimal=3)
else:
assert (qn.objective - 0.43780848383903503) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.5337948],
[1.678699],
[0.8060587]]),
decimal=3)
else:
if l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.120777],
[3.056865]]),
decimal=3)
else:
assert (qn.objective - 0.4750209450721741) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.3931049],
[2.0140104]]),
decimal=3)
if penalty == 'elasticnet':
if fit_intercept:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.1088872],
[2.4812558],
[0.7960136]]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.43780848383903503) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.5337948],
[1.678699],
[0.8060587]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.44295936822891235) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.6899368],
[1.9021575],
[0.8057671]]),
decimal=3)
else:
assert (qn.objective - 0.467987984418869) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.3727235],
[1.4639963],
[0.79312485]]),
decimal=3)
else:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-2.120777],
[3.056865]]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.4750209450721741) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.3931049],
[2.0140104]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.4769895672798157) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.6214856],
[2.3650239]]),
decimal=3)
else:
assert (qn.objective - 0.5067970156669617) < tol
np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
np.array([[-1.2102532],
[1.752459]]),
decimal=3)
print()
elif penalty == 'softmax':
pytest.skip("Better initial conditions for softmax tests are "
"in progress.")
# todo add tests for softmax dtype=np.float64
# elasticnet for this points converged to different solution
# if loss == 'softmax':
# if penalty == 'none' and l1_strength == 0.0 and l2_strength == 0.0:
# if fit_intercept:
# assert (qn.objective - 0.007433414924889803) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[15.236361,
# -41.595913,
# -33.55021],
# [-36.607555,
# -13.91267,
# -42.66093],
# [-25.04939,
# -26.793947,
# -31.50192]]),
# decimal=3)
# else:
# assert (qn.objective - 0.18794211745262146) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[14.2959795,
# -104.63812,
# -96.41866],
# [-105.31236,
# -170.4887,
# -96.486]]),
# decimal=3)
# elif penalty == 'l1' and l2_strength == 0.0:
# if fit_intercept:
# if l1_strength == 0.0:
# assert (qn.objective - 0.007433414924889803) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[15.236361,
# -41.595913,
# -33.55021],
# [-36.607555,
# -13.91267,
# -42.66093],
# [-25.04939,
# -26.793947,
# -31.50192]]),
# decimal=3)
# else:
# assert (qn.objective - 0.2925984263420105) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.2279763,
# -2.011927,
# -1.8038181],
# [-3.3828118,
# -0.64903206,
# -3.0688426],
# [-1.6962943,
# -0.8585775,
# -1.1564851]]),
# decimal=3)
# else:
# if l1_strength == 0.0:
# assert (qn.objective - 0.18794211745262146) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[14.2959795,
# -104.63812,
# -96.41866],
# [-105.31236,
# -170.4887,
# -96.486]]),
# decimal=3)
# else:
# assert (qn.objective - 0.3777262568473816) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.4765631,
# -1.569497,
# -0.6421711],
# [-2.0787644,
# -1.593922,
# -0.73674846]]),
# decimal=3)
# elif penalty == 'l2' and l1_strength == 0.0:
# if fit_intercept:
# if l2_strength == 0.0:
# assert (qn.objective - 0.007433414924889803) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[15.236361,
# -41.595913,
# -33.55021],
# [-36.607555,
# -13.91267,
# -42.66093],
# [-25.04939,
# -26.793947,
# -31.50192]]),
# decimal=3)
# else:
# assert (qn.objective - 0.28578639030456543) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.6702422,
# -1.5495867,
# -1.193351],
# [-2.207053,
# -0.6854614,
# -2.0305414],
# [-1.1746005,
# -0.7992407,
# -1.0034739]]),
# decimal=3)
# else:
# if l2_strength == 0.0:
# assert (qn.objective - 0.18794211745262146) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[14.2959795,
# -104.63812,
# -96.41866],
# [-105.31236,
# -170.4887,
# -96.486]]),
# decimal=3)
# else:
# assert (qn.objective - 0.3537392020225525) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.3769588,
# -1.0002015,
# -0.5205092],
# [-1.5185534,
# -1.029575,
# -0.47429192]]),
# decimal=3)
# if penalty == 'elasticnet':
# if fit_intercept:
# if l1_strength == 0.0 and l2_strength == 0.0:
# assert (qn.objective - 0.007433414924889803) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[15.236361,
# -41.595913,
# -33.55021],
# [-36.607555,
# -13.91267,
# -42.66093],
# [-25.04939,
# -26.793947,
# -31.50192]]),
# decimal=3)
# elif l1_strength == 0.0:
# assert (qn.objective - 0.28578639030456543) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.6702422,
# -1.5495867,
# -1.193351],
# [-2.207053,
# -0.6854614,
# -2.0305414],
# [-1.1746005,
# -0.7992407,
# -1.0034739]]),
# decimal=3)
# elif l2_strength == 0.0:
# assert (qn.objective - 0.2925984263420105) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.2279763,
# -2.011927,
# -1.8038181],
# [-3.3828118,
# -0.64903206,
# -3.0688426],
# [-1.6962943,
# -0.8585775,
# -1.1564851]]),
# decimal=3)
# else:
# assert (qn.objective - 0.34934690594673157) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.1901233,
# -1.2236115,
# -1.0416932],
# [-2.3100038,
# -0.46381754,
# -2.1544967],
# [-1.0984052,
# -0.44855425,
# -0.7347126]]),
# decimal=3)
# else:
# if l1_strength == 0.0 and l2_strength == 0.0:
# assert (qn.objective - 0.18794211745262146) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[14.2959795,
# -104.63812,
# -96.41866],
# [-105.31236,
# -170.4887,
# -96.486]]),
# decimal=3)
# elif l1_strength == 0.0:
# assert (qn.objective - 0.3537392020225525) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.3769588,
# -1.0002015,
# -0.5205092],
# [-1.5185534,
# -1.029575,
# -0.47429192]]),
# decimal=3)
# elif l2_strength == 0.0:
# assert (qn.objective - 0.3777262568473816) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.4765631,
# -1.569497,
# -0.6421711],
# [-2.0787644,
# -1.593922,
# -0.73674846]]),
# decimal=3)
# else:
# assert (qn.objective - 0.40656331181526184) < tol
# np.testing.assert_almost_equal(qn.coef_.copy_to_host(),
# np.array([[1.2176441,
# -0.8387626,
# -0.3155345],
# [-1.3095317,
# -0.60578823,
# -0.26777366]]),
# decimal=3)
if penalty == "none" and (l1_strength > 0 or l2_strength > 0):
pytest.skip("`none` penalty does not take l1/l2_strength")
def test_qn(loss, dtype, penalty, l1_strength, l2_strength, fit_intercept):
if penalty == "none" and (l1_strength > 0 or l2_strength > 0):
pytest.skip("`none` penalty does not take l1/l2_strength")
tol = 1e-6
X = np.array(precomputed_X, dtype=dtype)
if loss == 'sigmoid':
y = np.array(precomputed_y_log, dtype=dtype)
else:
y = np.array(precomputed_y_multi, dtype=dtype)
qn = cuQN(loss=loss, fit_intercept=fit_intercept, l1_strength=l1_strength,
l2_strength=l2_strength, tol=1e-8, output_type="cupy")
qn.fit(X, y)
print(qn.objective)
print(qn.coef_)
if loss == 'sigmoid':
if penalty == 'none' and l1_strength == 0.0 and l2_strength == 0.0:
if fit_intercept:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.120777],
[3.056865]]),
decimal=3)
elif penalty == 'l1' and l2_strength == 0.0:
if fit_intercept:
if l1_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.44295936822891235) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.6899368],
[1.9021575]]),
decimal=3)
else:
if l1_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.120777],
[3.056865]]),
decimal=3)
else:
assert (qn.objective - 0.4769895672798157) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.6214856],
[2.3650239]]),
decimal=3)
# assert False
elif penalty == 'l2' and l1_strength == 0.0:
if fit_intercept:
if l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.43780848383903503) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.5337948],
[1.678699]]),
decimal=3)
else:
if l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.120777],
[3.056865]]),
decimal=3)
else:
assert (qn.objective - 0.4750209450721741) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.3931049],
[2.0140104]]),
decimal=3)
if penalty == 'elasticnet':
if fit_intercept:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.1088872],
[2.4812558]]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.43780848383903503) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.5337948],
[1.678699]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.44295936822891235) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.6899368],
[1.9021575]]),
decimal=3)
else:
assert (qn.objective - 0.467987984418869) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.3727235],
[1.4639963]]),
decimal=3)
else:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-2.120777],
[3.056865]]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.4750209450721741) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.3931049],
[2.0140104]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.4769895672798157) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.6214856],
[2.3650239]]),
decimal=3)
else:
assert (qn.objective - 0.5067970156669617) < tol
cp.testing.assert_array_almost_equal(
qn.coef_,
np.array([[-1.2102532],
[1.752459]]),
decimal=3)
print()
elif loss == 'softmax':
pytest.skip("Better initial conditions for softmax tests are "
"in progress.")
def test_qn(loss, dtype, penalty, l1_strength, l2_strength, fit_intercept):
if penalty == "none" and (l1_strength > 0 or l2_strength > 0):
pytest.skip("`none` penalty does not take l1/l2_strength")
tol = 1e-6
qn = cuQN(loss=loss,
fit_intercept=fit_intercept,
l1_strength=l1_strength,
l2_strength=l2_strength,
tol=1e-8,
output_type="cupy")
if loss == 'softmax':
X, y = make_classification(n_samples=5000,
n_informative=10,
n_features=20,
n_classes=4,
dtype=dtype)
stratify = y.astype(dtype)
X_train, X_test, y_train, y_test = train_test_split(X.astype(dtype),
y.astype(dtype),
stratify=stratify)
most_class = cp.unique(y)[cp.argmax(cp.bincount(y))]
baseline_preds = cp.array([most_class] * y_test.shape[0], dtype=dtype)
baseline_score = accuracy_score(y_test, baseline_preds)
y_pred = qn.fit(X_train, y_train).predict(X_test)
cuml_score = accuracy_score(y_test, y_pred)
assert (cuml_score > baseline_score)
assert (cuml_score >= 0.50)
elif loss == 'sigmoid':
X = np.array(precomputed_X, dtype=dtype)
y = np.array(precomputed_y_log, dtype=dtype)
qn.fit(X, y)
print(qn.objective)
print(qn.coef_)
if penalty == 'none' and l1_strength == 0.0 and l2_strength == 0.0:
if fit_intercept:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array([[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array([[-2.120777],
[3.056865]]),
decimal=3)
elif penalty == 'l1' and l2_strength == 0.0:
if fit_intercept:
if l1_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.44295936822891235) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.6899368],
[1.9021575]]),
decimal=3)
else:
if l1_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array([[-2.120777],
[3.056865]
]),
decimal=3)
else:
assert (qn.objective - 0.4769895672798157) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.6214856],
[2.3650239]]),
decimal=3)
# assert False
elif penalty == 'l2' and l1_strength == 0.0:
if fit_intercept:
if l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-2.1088872],
[2.4812558]]),
decimal=3)
else:
assert (qn.objective - 0.43780848383903503) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.5337948],
[1.678699]]),
decimal=3)
else:
if l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array([[-2.120777],
[3.056865]
]),
decimal=3)
else:
assert (qn.objective - 0.4750209450721741) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.3931049],
[2.0140104]]),
decimal=3)
if penalty == 'elasticnet':
if fit_intercept:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.40263831615448) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-2.1088872],
[2.4812558]]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.43780848383903503) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.5337948],
[1.678699]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.44295936822891235) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.6899368],
[1.9021575]]),
decimal=3)
else:
assert (qn.objective - 0.467987984418869) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.3727235],
[1.4639963]]),
decimal=3)
else:
if l1_strength == 0.0 and l2_strength == 0.0:
assert (qn.objective - 0.4317452311515808) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array([[-2.120777],
[3.056865]
]),
decimal=3)
elif l1_strength == 0.0:
assert (qn.objective - 0.4750209450721741) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.3931049],
[2.0140104]]),
decimal=3)
elif l2_strength == 0.0:
assert (qn.objective - 0.4769895672798157) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.6214856],
[2.3650239]]),
decimal=3)
else:
assert (qn.objective - 0.5067970156669617) < tol
cp.testing.assert_array_almost_equal(qn.coef_,
np.array(
[[-1.2102532],
[1.752459]]),
decimal=3)
print()