scikit-learn-intelex 2025.4.0__py313-none-manylinux_2_28_x86_64.whl

sklearnex/_config.py

@@ -0,0 +1,116 @@
+# ==============================================================================
+# Copyright 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from contextlib import contextmanager
+from os import environ
+
+from sklearn import get_config as skl_get_config
+from sklearn import set_config as skl_set_config
+
+from daal4py.sklearn._utils import sklearn_check_version
+from onedal._config import _get_config as onedal_get_config
+
+
+def get_config():
+    """Retrieve current values for configuration set by :func:`set_config`
+    Returns
+    -------
+    config : dict
+        Keys are parameter names that can be passed to :func:`set_config`.
+    See Also
+    --------
+    config_context : Context manager for global configuration.
+    set_config : Set global configuration.
+    """
+    sklearn = skl_get_config()
+    sklearnex = onedal_get_config()
+    return {**sklearn, **sklearnex}
+
+
+def set_config(
+    target_offload=None,
+    allow_fallback_to_host=None,
+    allow_sklearn_after_onedal=None,
+    **sklearn_configs,
+):
+    """Set global configuration
+    Parameters
+    ----------
+    target_offload : string or dpctl.SyclQueue, default=None
+        The device primarily used to perform computations.
+        If string, expected to be "auto" (the execution context
+        is deduced from input data location),
+        or SYCL* filter selector string. Global default: "auto".
+    allow_fallback_to_host : bool, default=None
+        If True, allows to fallback computation to host device
+        in case particular estimator does not support the selected one.
+        Global default: False.
+    allow_sklearn_after_onedal : bool, default=None
+        If True, allows to fallback computation to sklearn after onedal
+        backend in case of runtime error on onedal backend computations.
+        Global default: True.
+    See Also
+    --------
+    config_context : Context manager for global configuration.
+    get_config : Retrieve current values of the global configuration.
+    """
+
+    array_api_dispatch = sklearn_configs.get("array_api_dispatch", False)
+    if array_api_dispatch and sklearn_check_version("1.6"):
+        environ["SCIPY_ARRAY_API"] = "1"
+
+    skl_set_config(**sklearn_configs)
+
+    local_config = onedal_get_config(copy=False)
+
+    if target_offload is not None:
+        local_config["target_offload"] = target_offload
+    if allow_fallback_to_host is not None:
+        local_config["allow_fallback_to_host"] = allow_fallback_to_host
+    if allow_sklearn_after_onedal is not None:
+        local_config["allow_sklearn_after_onedal"] = allow_sklearn_after_onedal
+
+
+@contextmanager
+def config_context(**new_config):
+    """Context manager for global scikit-learn configuration
+    Parameters
+    ----------
+    target_offload : string or dpctl.SyclQueue, default=None
+        The device primarily used to perform computations.
+        If string, expected to be "auto" (the execution context
+        is deduced from input data location),
+        or SYCL* filter selector string. Global default: "auto".
+    allow_fallback_to_host : bool, default=None
+        If True, allows to fallback computation to host device
+        in case particular estimator does not support the selected one.
+        Global default: False.
+    Notes
+    -----
+    All settings, not just those presently modified, will be returned to
+    their previous values when the context manager is exited.
+    See Also
+    --------
+    set_config : Set global scikit-learn configuration.
+    get_config : Retrieve current values of the global configuration.
+    """
+    old_config = get_config()
+    set_config(**new_config)
+
+    try:
+        yield
+    finally:
+        set_config(**old_config)

sklearnex/_device_offload.py

@@ -0,0 +1,126 @@
+# ==============================================================================
+# Copyright 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from functools import wraps
+
+from onedal._device_offload import _copy_to_usm, _get_global_queue, _transfer_to_host
+from onedal.utils._array_api import _asarray
+from onedal.utils._dpep_helpers import dpnp_available
+
+if dpnp_available:
+    import dpnp
+    from onedal.utils._array_api import _convert_to_dpnp
+
+from ._config import get_config
+
+
+def _get_backend(obj, queue, method_name, *data):
+    cpu_device = queue is None or queue.sycl_device.is_cpu
+    gpu_device = queue is not None and queue.sycl_device.is_gpu
+
+    if cpu_device:
+        patching_status = obj._onedal_cpu_supported(method_name, *data)
+        if patching_status.get_status():
+            return "onedal", queue, patching_status
+        else:
+            return "sklearn", None, patching_status
+
+    allow_fallback_to_host = get_config()["allow_fallback_to_host"]
+
+    if gpu_device:
+        patching_status = obj._onedal_gpu_supported(method_name, *data)
+        if patching_status.get_status():
+            return "onedal", queue, patching_status
+        else:
+            if allow_fallback_to_host:
+                patching_status = obj._onedal_cpu_supported(method_name, *data)
+                if patching_status.get_status():
+                    return "onedal", None, patching_status
+                else:
+                    return "sklearn", None, patching_status
+            else:
+                return "sklearn", None, patching_status
+
+    raise RuntimeError("Device support is not implemented")
+
+
+def dispatch(obj, method_name, branches, *args, **kwargs):
+    q = _get_global_queue()
+    has_usm_data_for_args, q, hostargs = _transfer_to_host(q, *args)
+    has_usm_data_for_kwargs, q, hostvalues = _transfer_to_host(q, *kwargs.values())
+    hostkwargs = dict(zip(kwargs.keys(), hostvalues))
+
+    backend, q, patching_status = _get_backend(obj, q, method_name, *hostargs)
+    has_usm_data = has_usm_data_for_args or has_usm_data_for_kwargs
+    if backend == "onedal":
+        # Host args only used before onedal backend call.
+        # Device will be offloaded when onedal backend will be called.
+        patching_status.write_log(queue=q, transferred_to_host=False)
+        return branches[backend](obj, *hostargs, **hostkwargs, queue=q)
+    if backend == "sklearn":
+        if (
+            "array_api_dispatch" in get_config()
+            and get_config()["array_api_dispatch"]
+            and "array_api_support" in obj._get_tags()
+            and obj._get_tags()["array_api_support"]
+            and not has_usm_data
+        ):
+            # USM ndarrays are also excluded for the fallback Array API. Currently, DPNP.ndarray is
+            # not compliant with the Array API standard, and DPCTL usm_ndarray Array API is compliant,
+            # except for the linalg module. There is no guarantee that stock scikit-learn will
+            # work with such input data. The condition will be updated after DPNP.ndarray and
+            # DPCTL usm_ndarray enabling for conformance testing and these arrays supportance
+            # of the fallback cases.
+            # If `array_api_dispatch` enabled and array api is supported for the stock scikit-learn,
+            # then raw inputs are used for the fallback.
+            patching_status.write_log(transferred_to_host=False)
+            return branches[backend](obj, *args, **kwargs)
+        else:
+            patching_status.write_log()
+            return branches[backend](obj, *hostargs, **hostkwargs)
+    raise RuntimeError(
+        f"Undefined backend {backend} in " f"{obj.__class__.__name__}.{method_name}"
+    )
+
+
+def wrap_output_data(func):
+    """
+    Converts and moves the output arrays of the decorated function
+    to match the input array type and device.
+    """
+
+    @wraps(func)
+    def wrapper(self, *args, **kwargs):
+        result = func(self, *args, **kwargs)
+        if not (len(args) == 0 and len(kwargs) == 0):
+            data = (*args, *kwargs.values())
+            usm_iface = getattr(data[0], "__sycl_usm_array_interface__", None)
+            if usm_iface is not None:
+                result = _copy_to_usm(usm_iface["syclobj"], result)
+                if dpnp_available and isinstance(data[0], dpnp.ndarray):
+                    result = _convert_to_dpnp(result)
+                return result
+            config = get_config()
+            if not ("transform_output" in config and config["transform_output"]):
+                input_array_api = getattr(data[0], "__array_namespace__", lambda: None)()
+                if input_array_api:
+                    input_array_api_device = data[0].device
+                    result = _asarray(
+                        result, input_array_api, device=input_array_api_device
+                    )
+        return result
+
+    return wrapper

sklearnex/_utils.py

@@ -0,0 +1,177 @@
+# ===============================================================================
+# Copyright 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ===============================================================================
+
+import logging
+import os
+import re
+import warnings
+from abc import ABC
+
+import sklearn
+
+from daal4py.sklearn._utils import (
+    PatchingConditionsChain as daal4py_PatchingConditionsChain,
+)
+from daal4py.sklearn._utils import daal_check_version, sklearn_check_version
+
+# Note: if inheriting from '_HTMLDocumentationLinkMixin' here, it then doesn't matter
+# the order of inheritance of classes for estimators when this is later subclassed,
+# whereas if inheriting from something else, the subclass that inherits from this needs
+# to be the first inherited class in estimators in order for it to take effect.
+if sklearn_check_version("1.4"):
+    from sklearn.utils._estimator_html_repr import _HTMLDocumentationLinkMixin
+
+    BaseForHTMLDocLink = _HTMLDocumentationLinkMixin
+else:
+    BaseForHTMLDocLink = ABC
+
+
+class PatchingConditionsChain(daal4py_PatchingConditionsChain):
+    def get_status(self):
+        return self.patching_is_enabled
+
+    def write_log(self, queue=None, transferred_to_host=True):
+        if self.patching_is_enabled:
+            self.logger.info(
+                f"{self.scope_name}: {get_patch_message('onedal', queue=queue, transferred_to_host=transferred_to_host)}"
+            )
+        else:
+            self.logger.debug(
+                f"{self.scope_name}: debugging for the patch is enabled to track"
+                " the usage of Intel® oneAPI Data Analytics Library (oneDAL)"
+            )
+            for message in self.messages:
+                self.logger.debug(
+                    f"{self.scope_name}: patching failed with cause - {message}"
+                )
+            self.logger.info(
+                f"{self.scope_name}: {get_patch_message('sklearn', transferred_to_host=transferred_to_host)}"
+            )
+
+
+def set_sklearn_ex_verbose():
+    log_level = os.environ.get("SKLEARNEX_VERBOSE")
+
+    logger = logging.getLogger("sklearnex")
+    logging_channel = logging.StreamHandler()
+    logging_formatter = logging.Formatter("%(levelname)s:%(name)s: %(message)s")
+    logging_channel.setFormatter(logging_formatter)
+    logger.addHandler(logging_channel)
+
+    try:
+        if log_level is not None:
+            logger.setLevel(log_level)
+    except Exception:
+        warnings.warn(
+            'Unknown level "{}" for logging.\n'
+            'Please, use one of "CRITICAL", "ERROR", '
+            '"WARNING", "INFO", "DEBUG".'.format(log_level)
+        )
+
+
+def get_patch_message(s, queue=None, transferred_to_host=True):
+    if s == "onedal":
+        message = "running accelerated version on "
+        if queue is not None:
+            if queue.sycl_device.is_gpu:
+                message += "GPU"
+            elif queue.sycl_device.is_cpu:
+                message += "CPU"
+            else:
+                raise RuntimeError("Unsupported device")
+        else:
+            message += "CPU"
+    elif s == "sklearn":
+        message = "fallback to original Scikit-learn"
+    elif s == "sklearn_after_onedal":
+        message = "failed to run accelerated version, fallback to original Scikit-learn"
+    else:
+        raise ValueError(
+            f"Invalid input - expected one of 'onedal','sklearn',"
+            f" 'sklearn_after_onedal', got {s}"
+        )
+    if transferred_to_host:
+        message += (
+            ". All input data transferred to host for further backend computations."
+        )
+    return message
+
+
+def get_sklearnex_version(rule):
+    return daal_check_version(rule)
+
+
+def register_hyperparameters(hyperparameters_map):
+    """Decorator for hyperparameters support in estimator class.
+    Adds `get_hyperparameters` method to class.
+    """
+
+    def decorator(cls):
+        """Add `get_hyperparameters()` static method"""
+
+        class StaticHyperparametersAccessor:
+            """Like a @staticmethod, but additionally raises a Warning when called on an instance."""
+
+            def __get__(self, instance, _):
+                if instance is not None:
+                    warnings.warn(
+                        "Hyperparameters are static variables and can not be modified per instance."
+                    )
+                return self.get_hyperparameters
+
+            def get_hyperparameters(self, op):
+                return hyperparameters_map[op]
+
+        cls.get_hyperparameters = StaticHyperparametersAccessor()
+        return cls
+
+    return decorator
+
+
+# This abstract class is meant to generate a clickable doc link for classses
+# in sklearnex that are not part of base scikit-learn.
+class IntelEstimator(BaseForHTMLDocLink):
+    @property
+    def _doc_link_module(self) -> str:
+        return "sklearnex"
+
+    @property
+    def _doc_link_template(self) -> str:
+        module_path, _ = self.__class__.__module__.rsplit(".", 1)
+        class_name = self.__class__.__name__
+        return f"https://uxlfoundation.github.io/scikit-learn-intelex/latest/non-scikit-algorithms.html#{module_path}.{class_name}"
+
+
+# This abstract class is meant to generate a clickable doc link for classses
+# in sklearnex that have counterparts in scikit-learn.
+class PatchableEstimator(BaseForHTMLDocLink):
+    @property
+    def _doc_link_module(self) -> str:
+        return "sklearnex"
+
+    @property
+    def _doc_link_template(self) -> str:
+        if re.search(r"^\d\.\d\.\d$", sklearn.__version__):
+            sklearn_version_parts = sklearn.__version__.split(".")
+            doc_version_url = sklearn_version_parts[0] + "." + sklearn_version_parts[1]
+        else:
+            doc_version_url = "stable"
+        module_path, _ = self.__class__.__module__.rsplit(".", 1)
+        module_path = re.sub("sklearnex", "sklearn", module_path)
+        class_name = self.__class__.__name__
+        # for TSNE, which re-uses daal4py
+        module_path = re.sub(r"daal4py\.", "", module_path)
+        return f"https://scikit-learn.org/{doc_version_url}/modules/generated/{module_path}.{class_name}.html"

sklearnex/basic_statistics/__init__.py

@@ -0,0 +1,20 @@
+# ==============================================================================
+# Copyright 2023 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from .basic_statistics import BasicStatistics
+from .incremental_basic_statistics import IncrementalBasicStatistics
+
+__all__ = ["BasicStatistics", "IncrementalBasicStatistics"]