scikit-learn-intelex 2024.6.0__py310-none-manylinux1_x86_64.whl → 2024.7.0__py310-none-manylinux1_x86_64.whl

sklearnex/ensemble/_forest.py

@@ -38,7 +38,12 @@ from sklearn.tree import (
 )
 from sklearn.tree._tree import Tree
 from sklearn.utils import check_random_state, deprecated
-from sklearn.utils.validation import check_array, check_is_fitted, check_X_y
+from sklearn.utils.validation import (
+    _check_sample_weight,
+    check_array,
+    check_is_fitted,
+    check_X_y,
+)
 
 from daal4py.sklearn._n_jobs_support import control_n_jobs
 from daal4py.sklearn._utils import (
@@ -70,7 +75,7 @@ class BaseForest(ABC):
         X, y = self._validate_data(
             X,
             y,
-            multi_output=False,
+            multi_output=True,
             accept_sparse=False,
             dtype=[np.float64, np.float32],
             force_all_finite=False,
@@ -78,7 +83,7 @@ class BaseForest(ABC):
         )
 
         if sample_weight is not None:
-            sample_weight = self.check_sample_weight(sample_weight, X)
+            sample_weight = _check_sample_weight(sample_weight, X)
 
         if y.ndim == 2 and y.shape[1] == 1:
             warnings.warn(
@@ -289,38 +294,6 @@ class BaseForest(ABC):
                 "min_bin_size must be integral number but was " "%r" % self.min_bin_size
             )
 
-    def check_sample_weight(self, sample_weight, X, dtype=None):
-        n_samples = _num_samples(X)
-
-        if dtype is not None and dtype not in [np.float32, np.float64]:
-            dtype = np.float64
-
-        if sample_weight is None:
-            sample_weight = np.ones(n_samples, dtype=dtype)
-        elif isinstance(sample_weight, numbers.Number):
-            sample_weight = np.full(n_samples, sample_weight, dtype=dtype)
-        else:
-            if dtype is None:
-                dtype = [np.float64, np.float32]
-            sample_weight = check_array(
-                sample_weight,
-                accept_sparse=False,
-                ensure_2d=False,
-                dtype=dtype,
-                order="C",
-                force_all_finite=False,
-            )
-            if sample_weight.ndim != 1:
-                raise ValueError("Sample weights must be 1D array or scalar")
-
-            if sample_weight.shape != (n_samples,):
-                raise ValueError(
-                    "sample_weight.shape == {}, expected {}!".format(
-                        sample_weight.shape, (n_samples,)
-                    )
-                )
-        return sample_weight
-
     @property
     def estimators_(self):
         if hasattr(self, "_cached_estimators_"):

sklearnex/ensemble/tests/test_forest.py

@@ -46,11 +46,10 @@ def test_sklearnex_import_rf_classifier(dataframe, queue):
     assert_allclose([1], _as_numpy(rf.predict([[0, 0, 0, 0]])))
 
 
-# TODO: fix RF regressor predict for the GPU sycl_queue.
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_rf_regression(dataframe, queue):
+    if queue and queue.sycl_device.is_gpu:
+        pytest.skip("RF regressor predict for the GPU sycl_queue is buggy.")
     from sklearnex.ensemble import RandomForestRegressor
 
     X, y = make_regression(n_features=4, n_informative=2, random_state=0, shuffle=False)
@@ -69,11 +68,10 @@ def test_sklearnex_import_rf_regression(dataframe, queue):
             assert_allclose([-6.839], pred, atol=1e-2)
 
 
-# TODO: fix ET classifier predict for the GPU sycl_queue.
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_et_classifier(dataframe, queue):
+    if queue and queue.sycl_device.is_gpu:
+        pytest.skip("ET classifier predict for the GPU sycl_queue is buggy.")
     from sklearnex.ensemble import ExtraTreesClassifier
 
     X, y = make_classification(
@@ -93,11 +91,10 @@ def test_sklearnex_import_et_classifier(dataframe, queue):
     assert_allclose([1], _as_numpy(rf.predict([[0, 0, 0, 0]])))
 
 
-# TODO: fix ET regressor predict for the GPU sycl_queue.
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_et_regression(dataframe, queue):
+    if queue and queue.sycl_device.is_gpu:
+        pytest.skip("ET regressor predict for the GPU sycl_queue is buggy.")
     from sklearnex.ensemble import ExtraTreesRegressor
 
     X, y = make_regression(n_features=1, random_state=0, shuffle=False)

sklearnex/linear_model/coordinate_descent.py

@@ -15,3 +15,16 @@
 # ===============================================================================
 
 from daal4py.sklearn.linear_model import ElasticNet, Lasso
+from onedal._device_offload import support_usm_ndarray
+
+# Note: `sklearnex.linear_model.ElasticNet` only has functional
+# sycl GPU support. No GPU device will be offloaded.
+ElasticNet.fit = support_usm_ndarray(queue_param=False)(ElasticNet.fit)
+ElasticNet.predict = support_usm_ndarray(queue_param=False)(ElasticNet.predict)
+ElasticNet.score = support_usm_ndarray(queue_param=False)(ElasticNet.score)
+
+# Note: `sklearnex.linear_model.Lasso` only has functional
+# sycl GPU support. No GPU device will be offloaded.
+Lasso.fit = support_usm_ndarray(queue_param=False)(Lasso.fit)
+Lasso.predict = support_usm_ndarray(queue_param=False)(Lasso.predict)
+Lasso.score = support_usm_ndarray(queue_param=False)(Lasso.score)

sklearnex/linear_model/linear.py

@@ -27,7 +27,6 @@ from daal4py.sklearn._utils import sklearn_check_version
 
 from .._device_offload import dispatch, wrap_output_data
 from .._utils import PatchingConditionsChain, get_patch_message, register_hyperparameters
-from ..utils.validation import _assert_all_finite
 
 if sklearn_check_version("1.0") and not sklearn_check_version("1.2"):
     from sklearn.linear_model._base import _deprecate_normalize
@@ -138,19 +137,6 @@ class LinearRegression(sklearn_LinearRegression):
             sample_weight=sample_weight,
         )
 
-    def _test_type_and_finiteness(self, X_in):
-        X = X_in if isinstance(X_in, np.ndarray) else np.asarray(X_in)
-
-        dtype = X.dtype
-        if "complex" in str(type(dtype)):
-            return False
-
-        try:
-            _assert_all_finite(X)
-        except BaseException:
-            return False
-        return True
-
     def _onedal_fit_supported(self, method_name, *data):
         assert method_name == "fit"
         assert len(data) == 3
@@ -174,7 +160,7 @@ class LinearRegression(sklearn_LinearRegression):
         # Check if equations are well defined
         is_underdetermined = n_samples < (n_features + int(self.fit_intercept))
 
-        dal_ready = patching_status.and_conditions(
+        patching_status.and_conditions(
             [
                 (sample_weight is None, "Sample weight is not supported."),
                 (
@@ -193,17 +179,6 @@ class LinearRegression(sklearn_LinearRegression):
                 ),
             ]
         )
-        if not dal_ready:
-            return patching_status
-
-        if not patching_status.and_condition(
-            self._test_type_and_finiteness(X), "Input X is not supported."
-        ):
-            return patching_status
-
-        patching_status.and_condition(
-            self._test_type_and_finiteness(y), "Input y is not supported."
-        )
 
         return patching_status
 
@@ -217,19 +192,13 @@ class LinearRegression(sklearn_LinearRegression):
         model_is_sparse = issparse(self.coef_) or (
             self.fit_intercept and issparse(self.intercept_)
         )
-        dal_ready = patching_status.and_conditions(
+        patching_status.and_conditions(
             [
                 (n_samples > 0, "Number of samples is less than 1."),
                 (not issparse(data[0]), "Sparse input is not supported."),
                 (not model_is_sparse, "Sparse coefficients are not supported."),
             ]
         )
-        if not dal_ready:
-            return patching_status
-
-        patching_status.and_condition(
-            self._test_type_and_finiteness(data[0]), "Input X is not supported."
-        )
 
         return patching_status
 
@@ -257,7 +226,6 @@ class LinearRegression(sklearn_LinearRegression):
             "accept_sparse": ["csr", "csc", "coo"],
             "y_numeric": True,
             "multi_output": True,
-            "force_all_finite": False,
         }
         if sklearn_check_version("1.2"):
             X, y = self._validate_data(**check_params)

sklearnex/linear_model/logistic_regression.py

@@ -28,7 +28,7 @@ if daal_check_version((2024, "P", 1)):
     from sklearn.linear_model import LogisticRegression as sklearn_LogisticRegression
     from sklearn.metrics import accuracy_score
     from sklearn.utils.multiclass import type_of_target
-    from sklearn.utils.validation import check_X_y
+    from sklearn.utils.validation import check_array, check_is_fitted, check_X_y
 
     from daal4py.sklearn._n_jobs_support import control_n_jobs
     from daal4py.sklearn._utils import sklearn_check_version
@@ -38,7 +38,8 @@ if daal_check_version((2024, "P", 1)):
 
     from .._device_offload import dispatch, wrap_output_data
     from .._utils import PatchingConditionsChain, get_patch_message
-    from ..utils.validation import _assert_all_finite
+
+    _sparsity_enabled = daal_check_version((2024, "P", 700))
 
     class BaseLogisticRegression(ABC):
         def _save_attributes(self):
@@ -107,8 +108,6 @@ if daal_check_version((2024, "P", 1)):
         _onedal_cpu_fit = daal4py_fit
 
         def fit(self, X, y, sample_weight=None):
-            if sklearn_check_version("1.0"):
-                self._check_feature_names(X, reset=True)
             if sklearn_check_version("1.2"):
                 self._validate_params()
             dispatch(
@@ -126,8 +125,6 @@ if daal_check_version((2024, "P", 1)):
 
         @wrap_output_data
         def predict(self, X):
-            if sklearn_check_version("1.0"):
-                self._check_feature_names(X, reset=False)
             return dispatch(
                 self,
                 "predict",
@@ -140,8 +137,6 @@ if daal_check_version((2024, "P", 1)):
 
         @wrap_output_data
         def predict_proba(self, X):
-            if sklearn_check_version("1.0"):
-                self._check_feature_names(X, reset=False)
             return dispatch(
                 self,
                 "predict_proba",
@@ -154,8 +149,6 @@ if daal_check_version((2024, "P", 1)):
 
         @wrap_output_data
         def predict_log_proba(self, X):
-            if sklearn_check_version("1.0"):
-                self._check_feature_names(X, reset=False)
             return dispatch(
                 self,
                 "predict_log_proba",
@@ -168,8 +161,6 @@ if daal_check_version((2024, "P", 1)):
 
         @wrap_output_data
         def score(self, X, y, sample_weight=None):
-            if sklearn_check_version("1.0"):
-                self._check_feature_names(X, reset=False)
             return dispatch(
                 self,
                 "score",
@@ -187,17 +178,6 @@ if daal_check_version((2024, "P", 1)):
                 y, self._onedal_predict(X, queue=queue), sample_weight=sample_weight
             )
 
-        def _test_type_and_finiteness(self, X_in):
-            X = np.asarray(X_in)
-
-            if np.iscomplexobj(X):
-                return False
-            try:
-                _assert_all_finite(X)
-            except BaseException:
-                return False
-            return True
-
         def _onedal_gpu_fit_supported(self, method_name, *data):
             assert method_name == "fit"
             assert len(data) == 3
@@ -208,7 +188,12 @@ if daal_check_version((2024, "P", 1)):
                 f"sklearn.linear_model.{class_name}.fit"
             )
 
-            dal_ready = patching_status.and_conditions(
+            target_type = (
+                type_of_target(y, input_name="y")
+                if sklearn_check_version("1.1")
+                else type_of_target(y)
+            )
+            patching_status.and_conditions(
                 [
                     (self.penalty == "l2", "Only l2 penalty is supported."),
                     (self.dual == False, "dual=True is not supported."),
@@ -226,24 +211,12 @@ if daal_check_version((2024, "P", 1)):
                     (self.l1_ratio is None, "l1 ratio is not supported."),
                     (sample_weight is None, "Sample weight is not supported."),
                     (
-                        type_of_target(y) == "binary",
+                        target_type == "binary",
                         "Only binary classification is supported",
                     ),
                 ]
             )
 
-            if not dal_ready:
-                return patching_status
-
-            if not patching_status.and_condition(
-                self._test_type_and_finiteness(X), "Input X is not supported."
-            ):
-                return patching_status
-
-            patching_status.and_condition(
-                self._test_type_and_finiteness(y), "Input y is not supported."
-            )
-
             return patching_status
 
         def _onedal_gpu_predict_supported(self, method_name, *data):
@@ -267,7 +240,7 @@ if daal_check_version((2024, "P", 1)):
                 [
                     (n_samples > 0, "Number of samples is less than 1."),
                     (
-                        not any([issparse(i) for i in data]),
+                        (not any([issparse(i) for i in data])) or _sparsity_enabled,
                         "Sparse input is not supported.",
                     ),
                     (not model_is_sparse, "Sparse coefficients are not supported."),
@@ -277,12 +250,6 @@ if daal_check_version((2024, "P", 1)):
                     ),
                 ]
             )
-            if not dal_ready:
-                return patching_status
-
-            patching_status.and_condition(
-                self._test_type_and_finiteness(*data), "Input X is not supported."
-            )
 
             return patching_status
 
@@ -313,24 +280,29 @@ if daal_check_version((2024, "P", 1)):
             }
             self._onedal_estimator = onedal_LogisticRegression(**onedal_params)
 
-        def _onedal_fit(self, X, y, sample_weight, queue=None):
+        def _onedal_fit(self, X, y, sample_weight=None, queue=None):
             if queue is None or queue.sycl_device.is_cpu:
                 return self._onedal_cpu_fit(X, y, sample_weight)
 
             assert sample_weight is None
 
-            check_params = {
-                "X": X,
-                "y": y,
-                "dtype": [np.float64, np.float32],
-                "accept_sparse": False,
-                "multi_output": False,
-                "force_all_finite": True,
-            }
-            if sklearn_check_version("1.2"):
-                X, y = self._validate_data(**check_params)
+            if sklearn_check_version("1.0"):
+                X, y = self._validate_data(
+                    X,
+                    y,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
             else:
-                X, y = check_X_y(**check_params)
+                X, y = check_X_y(
+                    X,
+                    y,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+
             self._initialize_onedal_estimator()
             try:
                 self._onedal_estimator.fit(X, y, queue=queue)
@@ -348,7 +320,23 @@ if daal_check_version((2024, "P", 1)):
             if queue is None or queue.sycl_device.is_cpu:
                 return daal4py_predict(self, X, "computeClassLabels")
 
-            X = self._validate_data(X, accept_sparse=False, reset=False)
+            check_is_fitted(self)
+            if sklearn_check_version("1.0"):
+                X = self._validate_data(
+                    X,
+                    reset=False,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+            else:
+                X = check_array(
+                    X,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+
             assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict(X, queue=queue)
 
@@ -356,7 +344,23 @@ if daal_check_version((2024, "P", 1)):
             if queue is None or queue.sycl_device.is_cpu:
                 return daal4py_predict(self, X, "computeClassProbabilities")
 
-            X = self._validate_data(X, accept_sparse=False, reset=False)
+            check_is_fitted(self)
+            if sklearn_check_version("1.0"):
+                X = self._validate_data(
+                    X,
+                    reset=False,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+            else:
+                X = check_array(
+                    X,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+
             assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict_proba(X, queue=queue)
 
@@ -364,7 +368,23 @@ if daal_check_version((2024, "P", 1)):
             if queue is None or queue.sycl_device.is_cpu:
                 return daal4py_predict(self, X, "computeClassLogProbabilities")
 
-            X = self._validate_data(X, accept_sparse=False, reset=False)
+            check_is_fitted(self)
+            if sklearn_check_version("1.0"):
+                X = self._validate_data(
+                    X,
+                    reset=False,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+            else:
+                X = check_array(
+                    X,
+                    accept_sparse=_sparsity_enabled,
+                    accept_large_sparse=_sparsity_enabled,
+                    dtype=[np.float64, np.float32],
+                )
+
             assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict_log_proba(X, queue=queue)
 

sklearnex/linear_model/ridge.py

@@ -15,3 +15,10 @@
 # ===============================================================================
 
 from daal4py.sklearn.linear_model import Ridge
+from onedal._device_offload import support_usm_ndarray
+
+# Note: `sklearnex.linear_model.Ridge` only has functional
+# sycl GPU support. No GPU device will be offloaded.
+Ridge.fit = support_usm_ndarray(queue_param=False)(Ridge.fit)
+Ridge.predict = support_usm_ndarray(queue_param=False)(Ridge.predict)
+Ridge.score = support_usm_ndarray(queue_param=False)(Ridge.score)

sklearnex/linear_model/tests/test_linear.py

@@ -20,6 +20,10 @@ from numpy.testing import assert_allclose
 from sklearn.datasets import make_regression
 
 from daal4py.sklearn._utils import daal_check_version
+from daal4py.sklearn.linear_model.tests.test_ridge import (
+    _test_multivariate_ridge_alpha_shape,
+    _test_multivariate_ridge_coefficients,
+)
 from onedal.tests.utils._dataframes_support import (
     _as_numpy,
     _convert_to_dataframe,
@@ -57,32 +61,41 @@ def test_sklearnex_import_linear(dataframe, queue, dtype, macro_block):
     assert_allclose(_as_numpy(linreg.coef_), [1.0, 2.0], rtol=tol)
 
 
-def test_sklearnex_import_ridge():
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+def test_sklearnex_import_ridge(dataframe, queue):
     from sklearnex.linear_model import Ridge
 
     X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
     y = np.dot(X, np.array([1, 2])) + 3
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    y = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
     ridgereg = Ridge().fit(X, y)
     assert "daal4py" in ridgereg.__module__
     assert_allclose(ridgereg.intercept_, 4.5)
     assert_allclose(ridgereg.coef_, [0.8, 1.4])
 
 
-def test_sklearnex_import_lasso():
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+def test_sklearnex_import_lasso(dataframe, queue):
     from sklearnex.linear_model import Lasso
 
     X = [[0, 0], [1, 1], [2, 2]]
     y = [0, 1, 2]
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    y = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
     lasso = Lasso(alpha=0.1).fit(X, y)
     assert "daal4py" in lasso.__module__
     assert_allclose(lasso.intercept_, 0.15)
     assert_allclose(lasso.coef_, [0.85, 0.0])
 
 
-def test_sklearnex_import_elastic():
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+def test_sklearnex_import_elastic(dataframe, queue):
     from sklearnex.linear_model import ElasticNet
 
     X, y = make_regression(n_features=2, random_state=0)
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    y = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
     elasticnet = ElasticNet(random_state=0).fit(X, y)
     assert "daal4py" in elasticnet.__module__
     assert_allclose(elasticnet.intercept_, 1.451, atol=1e-3)
@@ -115,3 +128,15 @@ def test_sklearnex_reconstruct_model(dataframe, queue, dtype):
 
     tol = 1e-5 if _as_numpy(y_pred).dtype == np.float32 else 1e-7
     assert_allclose(gtr, _as_numpy(y_pred), rtol=tol)
+
+
+def test_sklearnex_multivariate_ridge_coefs():
+    from sklearnex.linear_model import Ridge
+
+    _test_multivariate_ridge_coefficients(Ridge, random_state=0)
+
+
+def test_sklearnex_multivariate_ridge_alpha_shape():
+    from sklearnex.linear_model import Ridge
+
+    _test_multivariate_ridge_alpha_shape(Ridge, random_state=0)

sklearnex/linear_model/tests/test_logreg.py

@@ -14,8 +14,11 @@
 # limitations under the License.
 # ===============================================================================
 
+import numpy as np
 import pytest
-from sklearn.datasets import load_breast_cancer, load_iris
+from numpy.testing import assert_allclose, assert_array_equal
+from scipy.sparse import csr_matrix
+from sklearn.datasets import load_breast_cancer, load_iris, make_classification
 from sklearn.metrics import accuracy_score
 from sklearn.model_selection import train_test_split
 
@@ -24,7 +27,9 @@ from onedal.tests.utils._dataframes_support import (
     _as_numpy,
     _convert_to_dataframe,
     get_dataframes_and_queues,
+    get_queues,
 )
+from sklearnex import config_context
 
 
 def prepare_input(X, y, dataframe, queue):
@@ -38,8 +43,7 @@ def prepare_input(X, y, dataframe, queue):
 
 
 @pytest.mark.parametrize(
-    "dataframe,queue",
-    get_dataframes_and_queues(device_filter_="cpu"),
+    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
 )
 def test_sklearnex_multiclass_classification(dataframe, queue):
     from sklearnex.linear_model import LogisticRegression
@@ -89,3 +93,41 @@ def test_sklearnex_binary_classification(dataframe, queue):
 
     y_pred = _as_numpy(logreg.predict(X_test))
     assert accuracy_score(y_test, y_pred) > 0.95
+
+
+if daal_check_version((2024, "P", 700)):
+
+    @pytest.mark.parametrize("queue", get_queues("gpu"))
+    @pytest.mark.parametrize("dtype", [np.float32, np.float64])
+    @pytest.mark.parametrize(
+        "dims", [(3007, 17, 0.05), (50000, 100, 0.01), (512, 10, 0.5)]
+    )
+    def test_csr(queue, dtype, dims):
+        from sklearnex.linear_model import LogisticRegression
+
+        n, p, density = dims
+
+        # Create sparse dataset for classification
+        X, y = make_classification(n, p, random_state=42)
+        X = X.astype(dtype)
+        y = y.astype(dtype)
+        np.random.seed(2007 + n + p)
+        mask = np.random.binomial(1, density, (n, p))
+        X = X * mask
+        X_sp = csr_matrix(X)
+
+        model = LogisticRegression(fit_intercept=True, solver="newton-cg")
+        model_sp = LogisticRegression(fit_intercept=True, solver="newton-cg")
+
+        with config_context(target_offload="gpu:0"):
+            model.fit(X, y)
+            pred = model.predict(X)
+            prob = model.predict_proba(X)
+            model_sp.fit(X_sp, y)
+            pred_sp = model_sp.predict(X_sp)
+            prob_sp = model_sp.predict_proba(X_sp)
+
+        assert_allclose(pred, pred_sp)
+        assert_allclose(prob, prob_sp)
+        assert_allclose(model.coef_, model_sp.coef_, rtol=1e-4)
+        assert_allclose(model.intercept_, model_sp.intercept_, rtol=1e-4)

sklearnex/manifold/t_sne.py

@@ -15,3 +15,7 @@
 # ===============================================================================
 
 from daal4py.sklearn.manifold import TSNE
+from onedal._device_offload import support_usm_ndarray
+
+TSNE.fit = support_usm_ndarray(queue_param=False)(TSNE.fit)
+TSNE.fit_transform = support_usm_ndarray(queue_param=False)(TSNE.fit_transform)

sklearnex/metrics/pairwise.py

@@ -15,3 +15,8 @@
 # ===============================================================================
 
 from daal4py.sklearn.metrics import pairwise_distances
+from onedal._device_offload import support_usm_ndarray
+
+pairwise_distances = support_usm_ndarray(freefunc=True, queue_param=False)(
+    pairwise_distances
+)

sklearnex/metrics/ranking.py

@@ -15,3 +15,6 @@
 # ===============================================================================
 
 from daal4py.sklearn.metrics import roc_auc_score
+from onedal._device_offload import support_usm_ndarray
+
+roc_auc_score = support_usm_ndarray(freefunc=True, queue_param=False)(roc_auc_score)

sklearnex/model_selection/split.py

@@ -15,3 +15,6 @@
 # ===============================================================================
 
 from daal4py.sklearn.model_selection import train_test_split
+from onedal._device_offload import support_usm_ndarray
+
+train_test_split = support_usm_ndarray(freefunc=True, queue_param=False)(train_test_split)

sklearnex/neighbors/_lof.py

@@ -97,6 +97,15 @@ class LocalOutlierFactor(KNeighborsDispatchingBase, sklearn_LocalOutlierFactor):
                 self.negative_outlier_factor_, 100.0 * self.contamination
             )
 
+        # adoption of warning for data with duplicated samples from
+        # https://github.com/scikit-learn/scikit-learn/pull/28773
+        if sklearn_check_version("1.6"):
+            if np.min(self.negative_outlier_factor_) < -1e7 and not self.novelty:
+                warnings.warn(
+                    "Duplicate values are leading to incorrect results. "
+                    "Increase the number of neighbors for more accurate results."
+                )
+
         return self
 
     def fit(self, X, y=None):