update user guide

tvdboom · Aug 28, 2023 · 310735a · 310735a
1 parent b38e87b
commit 310735a
Show file tree

Hide file tree

Showing 211 changed files with 5,235 additions and 7,367 deletions.
diff --git a/atom/api.py b/atom/api.py
@@ -222,10 +222,10 @@ class ATOMClassifier(BaseTransformer, ATOM):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -448,10 +448,10 @@ class ATOMForecaster(BaseTransformer, ATOM):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -677,10 +677,10 @@ class ATOMRegressor(BaseTransformer, ATOM):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and

diff --git a/atom/basemodel.py b/atom/basemodel.py
@@ -102,10 +102,10 @@ class BaseModel(BaseTransformer, BaseTracker, HTPlot, PredictionPlot, ShapPlot):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and

diff --git a/atom/basetransformer.py b/atom/basetransformer.py
@@ -50,7 +50,7 @@ class BaseTransformer:
         Standard keyword arguments for the classes. Can include:
 
         - n_jobs: Number of cores to use for parallel processing.
-        - device: Device on which to train the estimators.
+        - device: Device on which to run the estimators.
         - engine: Execution engine to use for data and estimators.
         - backend: Parallelization backend.
         - verbose: Verbosity level of the output.
@@ -103,13 +103,13 @@ def n_jobs(self, value: INT):
 
     @property
     def device(self) -> str:
-        """Device on which to train the estimators."""
+        """Device on which to run the estimators."""
         return self._device
 
     @device.setter
     def device(self, value: str):
-        self._device = value
-        if "gpu" in value.lower():
+        self._device = value.lower()
+        if "gpu" in self._device:
             os.environ["CUDA_VISIBLE_DEVICES"] = str(self._device_id)
 
     @property
@@ -155,16 +155,21 @@ def engine(self, value: dict | None):
                     )
                 else:
                     import sklearnex
-                    sklearnex.set_config("auto" if "cpu" in self.device else self.device)
+                    sklearnex.set_config(self.device if "gpu" in self.device else "auto")
             elif models.lower() == "cuml":
                 if not find_spec("cuml"):
                     raise ModuleNotFoundError(
                         "Failed to import cuml. Package is not installed. Refer "
                         "to: https://rapids.ai/start.html#install."
                     )
                 else:
-                    import cuml
-                    cuml.internals.memory_utils.set_global_output_type("numpy")
+                    from cuml.common.device_selection import set_global_device_type
+                    set_global_device_type("gpu" if "gpu" in self.device else "cpu")
+
+                    # See https://github.com/rapidsai/cuml/issues/5564
+                    from cuml.internals.memory_util import set_global_output_type
+                    set_global_output_type("numpy")
+
             elif models.lower() != "sklearn":
                 raise ValueError(
                     "Invalid value for the models key of the engine parameter, "

diff --git a/atom/data_cleaning.py b/atom/data_cleaning.py
@@ -539,10 +539,10 @@ class Cleaner(BaseEstimator, TransformerMixin, BaseTransformer):
         is not provided.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -972,10 +972,10 @@ class Discretizer(BaseEstimator, TransformerMixin, BaseTransformer):
         - If dict: Labels per column, where the key is the column's name.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1694,10 +1694,10 @@ class Imputer(BaseEstimator, TransformerMixin, BaseTransformer):
         (if more, the column is removed). If None, ignore this step.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -2095,10 +2095,10 @@ class Normalizer(BaseEstimator, TransformerMixin, BaseTransformer):
         - "[quantile][]": Transform features using quantiles information.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -2394,10 +2394,10 @@ class Pruner(BaseEstimator, TransformerMixin, BaseTransformer):
         if strategy="zscore".
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -2699,10 +2699,10 @@ class Scaler(BaseEstimator, TransformerMixin, BaseTransformer):
         Whether to scale binary columns (only 0s and 1s).
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and

diff --git a/atom/feature_engineering.py b/atom/feature_engineering.py
@@ -991,10 +991,10 @@ class FeatureSelector(
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and

diff --git a/atom/models.py b/atom/models.py
@@ -499,7 +499,7 @@ class CatBoost(ClassRegModel):
           `use_best_model=False` parameter to avoid this behavior or use
           a [holdout set][data-sets] to evaluate the final estimator.
         * [In-training validation][] and [pruning][] are disabled when
-          `device="gpu"`.
+          `#!python device="gpu"`.
 
     !!! note
         ATOM uses CatBoost's `n_estimators` parameter instead of
@@ -1757,8 +1757,8 @@ class LightGBM(ClassRegModel):
     Read more in LightGBM's [documentation][lgbdocs].
 
     !!! info
-        Using LightGBM's [GPU acceleration][] requires
-        [additional software dependencies][lgb_gpu].
+        Using LightGBM's [GPU acceleration][estimator-acceleration]
+        requires [additional software dependencies][lgb_gpu].
 
     See Also
     --------

diff --git a/atom/nlp.py b/atom/nlp.py
@@ -831,10 +831,10 @@ class Vectorizer(BaseEstimator, TransformerMixin, BaseTransformer):
         in X (besides `corpus`) that are non-sparse.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and

diff --git a/atom/training.py b/atom/training.py
@@ -345,10 +345,10 @@ class DirectClassifier(Direct):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -563,10 +563,10 @@ class DirectForecaster(Direct):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -778,10 +778,10 @@ class DirectRegressor(Direct):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1003,10 +1003,10 @@ class SuccessiveHalvingClassifier(SuccessiveHalving):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1225,10 +1225,10 @@ class SuccessiveHalvingForecaster(SuccessiveHalving):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1444,10 +1444,10 @@ class SuccessiveHalvingRegressor(SuccessiveHalving):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1675,10 +1675,10 @@ class TrainSizingClassifier(TrainSizing):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -1902,10 +1902,10 @@ class TrainSizingForecaster(TrainSizing):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and
@@ -2126,10 +2126,10 @@ class TrainSizingRegressor(TrainSizing):
         - If <-1: Use number of cores - 1 + `n_jobs`.
 
     device: str, default="cpu"
-        Device on which to train the estimators. Use any string
-        that follows the [SYCL_DEVICE_FILTER][] filter selector,
-        e.g. `device="gpu"` to use the GPU. Read more in the
-        [user guide][accelerating-pipelines].
+        Device on which to run the estimators. Use any string that
+        follows the [SYCL_DEVICE_FILTER][] filter selector, e.g.
+        `#!python device="gpu"` to use the GPU. Read more in the
+        [user guide][gpu-acceleration].
 
     engine: dict or None, default=None
         Execution engine to use for [data][data-acceleration] and