Fix a few lingering sklearn compat xfails

python/cuml/cuml/ensemble/randomforest_common.pyx

@@ -322,7 +322,6 @@ class BaseRandomForestModel(Base, InteropMixin):
         min_impurity_decrease=0.0,
         max_batch_size=4096,
         random_state=None,
-        criterion=None,
         n_streams=4,
         oob_score=False,
         verbose=False,
@@ -644,8 +643,8 @@ class BaseRandomForestModel(Base, InteropMixin):
         )
         self.n_outputs_ = 1
         self._treelite_model_bytes = <bytes>(tl_bytes[:tl_bytes_len])
-        # Ensure cached nvforest model is reset
-        self._nvforest_model = None
+        # Reload nvforest model
+        self._nvforest_model = self.as_nvforest()
 
         # Compute OOB score if requested
         if self.oob_score:

python/cuml/cuml/ensemble/randomforestclassifier.py

@@ -190,16 +190,42 @@ def _attrs_to_cpu(self, model):
     def __init__(
         self,
         *,
+        n_estimators=100,
         split_criterion="gini",
+        bootstrap=True,
+        max_samples=1.0,
+        max_depth="deprecated",
+        max_leaves=-1,
+        max_features="sqrt",
+        n_bins=128,
+        min_samples_leaf=1,
+        min_samples_split=2,
+        min_impurity_decrease=0.0,
+        max_batch_size=4096,
+        random_state=None,
+        n_streams=4,
+        oob_score=False,
         verbose=False,
         output_type=None,
-        **kwargs,
     ):
         super().__init__(
             split_criterion=split_criterion,
+            n_estimators=n_estimators,
+            bootstrap=bootstrap,
+            max_samples=max_samples,
+            max_depth=max_depth,
+            max_leaves=max_leaves,
+            max_features=max_features,
+            n_bins=n_bins,
+            min_samples_leaf=min_samples_leaf,
+            min_samples_split=min_samples_split,
+            min_impurity_decrease=min_impurity_decrease,
+            max_batch_size=max_batch_size,
+            random_state=random_state,
+            n_streams=n_streams,
+            oob_score=oob_score,
             verbose=verbose,
             output_type=output_type,
-            **kwargs,
         )
 
     @nvtx.annotate(

python/cuml/cuml/ensemble/randomforestregressor.py

@@ -160,18 +160,42 @@ class RandomForestRegressor(BaseRandomForestModel, RegressorMixin):
     def __init__(
         self,
         *,
+        n_estimators=100,
         split_criterion="mse",
+        bootstrap=True,
+        max_samples=1.0,
+        max_depth="deprecated",
+        max_leaves=-1,
         max_features=1.0,
+        n_bins=128,
+        min_samples_leaf=1,
+        min_samples_split=2,
+        min_impurity_decrease=0.0,
+        max_batch_size=4096,
+        random_state=None,
+        n_streams=4,
+        oob_score=False,
         verbose=False,
         output_type=None,
-        **kwargs,
     ):
         super().__init__(
+            n_estimators=n_estimators,
             split_criterion=split_criterion,
+            bootstrap=bootstrap,
+            max_samples=max_samples,
+            max_depth=max_depth,
+            max_leaves=max_leaves,
             max_features=max_features,
+            n_bins=n_bins,
+            min_samples_leaf=min_samples_leaf,
+            min_samples_split=min_samples_split,
+            min_impurity_decrease=min_impurity_decrease,
+            max_batch_size=max_batch_size,
+            random_state=random_state,
+            n_streams=n_streams,
+            oob_score=oob_score,
             verbose=verbose,
             output_type=output_type,
-            **kwargs,
         )
 
     @nvtx.annotate(

python/cuml/cuml/neighbors/kneighbors_classifier.pyx

@@ -74,6 +74,36 @@ class KNeighborsClassifier(ClassifierMixin, FMajorInputTagMixin, NeighborsBase):
         - [callable] : a user-defined function which accepts an
           array of distances, and returns an array of the same shape
           containing the weights.
+    p : float (default=2)
+        Parameter for the Minkowski metric. When p = 1, this is equivalent to
+        manhattan distance (l1), and euclidean distance (l2) for p = 2. For
+        arbitrary p, minkowski distance (lp) is used.
+    algo_params : dict, optional (default=None)
+        Used to configure the nearest neighbor algorithm to be used.
+        If set to None, parameters will be generated automatically.
+        Parameters for algorithm ``'brute'`` when inputs are sparse:
+
+            - batch_size_index : (int) number of rows in each batch of \
+                                 index array
+            - batch_size_query : (int) number of rows in each batch of \
+                                 query array
+
+        Parameters for algorithm ``'ivfflat'``:
+
+            - nlist: (int) number of cells to partition dataset into
+            - nprobe: (int) at query time, number of cells used for search
+
+        Parameters for algorithm ``'ivfpq'``:
+
+            - nlist: (int) number of cells to partition dataset into
+            - nprobe: (int) at query time, number of cells used for search
+            - M: (int) number of subquantizers
+            - n_bits: (int) bits allocated per subquantizer
+            - usePrecomputedTables : (bool) whether to use precomputed tables
+    metric_params : dict, optional (default = None)
+        Additional keyword arguments for the metric function.
+    n_jobs : int (default = None)
+        Ignored, here for scikit-learn API compatibility.
     verbose : int or boolean, default=False
         Sets logging level. It must be one of `cuml.common.logger.level_*`.
         See :ref:`verbosity-levels` for more info.
@@ -154,12 +184,28 @@ class KNeighborsClassifier(ClassifierMixin, FMajorInputTagMixin, NeighborsBase):
     def __init__(
         self,
         *,
+        n_neighbors=5,
+        algorithm="auto",
+        metric="euclidean",
         weights="uniform",
+        p=2,
+        algo_params=None,
+        metric_params=None,
+        n_jobs=None,  # Ignored, here for sklearn API compatibility
         verbose=False,
         output_type=None,
-        **kwargs,
     ):
-        super().__init__(verbose=verbose, output_type=output_type, **kwargs)
+        super().__init__(
+            n_neighbors=n_neighbors,
+            algorithm=algorithm,
+            metric=metric,
+            p=p,
+            algo_params=algo_params,
+            metric_params=metric_params,
+            n_jobs=n_jobs,
+            verbose=verbose,
+            output_type=output_type,
+        )
         self.weights = weights
 
     @generate_docstring(convert_dtype_cast='np.float32')

python/cuml/cuml/neighbors/kneighbors_regressor.pyx

@@ -76,6 +76,36 @@ class KNeighborsRegressor(RegressorMixin, FMajorInputTagMixin, NeighborsBase):
         - [callable] : a user-defined function which accepts an
           array of distances, and returns an array of the same shape
           containing the weights.
+    p : float (default=2)
+        Parameter for the Minkowski metric. When p = 1, this is equivalent to
+        manhattan distance (l1), and euclidean distance (l2) for p = 2. For
+        arbitrary p, minkowski distance (lp) is used.
+    algo_params : dict, optional (default=None)
+        Used to configure the nearest neighbor algorithm to be used.
+        If set to None, parameters will be generated automatically.
+        Parameters for algorithm ``'brute'`` when inputs are sparse:
+
+            - batch_size_index : (int) number of rows in each batch of \
+                                 index array
+            - batch_size_query : (int) number of rows in each batch of \
+                                 query array
+
+        Parameters for algorithm ``'ivfflat'``:
+
+            - nlist: (int) number of cells to partition dataset into
+            - nprobe: (int) at query time, number of cells used for search
+
+        Parameters for algorithm ``'ivfpq'``:
+
+            - nlist: (int) number of cells to partition dataset into
+            - nprobe: (int) at query time, number of cells used for search
+            - M: (int) number of subquantizers
+            - n_bits: (int) bits allocated per subquantizer
+            - usePrecomputedTables : (bool) whether to use precomputed tables
+    metric_params : dict, optional (default = None)
+        Additional keyword arguments for the metric function.
+    n_jobs : int (default = None)
+        Ignored, here for scikit-learn API compatibility.
     verbose : int or boolean, default=False
         Sets logging level. It must be one of `cuml.common.logger.level_*`.
         See :ref:`verbosity-levels` for more info.
@@ -156,12 +186,28 @@ class KNeighborsRegressor(RegressorMixin, FMajorInputTagMixin, NeighborsBase):
     def __init__(
         self,
         *,
+        n_neighbors=5,
+        algorithm="auto",
+        metric="euclidean",
         weights="uniform",
+        p=2,
+        algo_params=None,
+        metric_params=None,
+        n_jobs=None,  # Ignored, here for sklearn API compatibility
         verbose=False,
         output_type=None,
-        **kwargs,
     ):
-        super().__init__(verbose=verbose, output_type=output_type, **kwargs)
+        super().__init__(
+            n_neighbors=n_neighbors,
+            algorithm=algorithm,
+            metric=metric,
+            p=p,
+            algo_params=algo_params,
+            metric_params=metric_params,
+            n_jobs=n_jobs,
+            verbose=verbose,
+            output_type=output_type,
+        )
         self.weights = weights
 
     @generate_docstring(convert_dtype_cast='np.float32')

python/cuml/cuml/neighbors/nearest_neighbors.pyx

@@ -559,13 +559,13 @@ class NeighborsBase(Base, InteropMixin, CMajorInputTagMixin, SparseInputTagMixin
         self,
         *,
         n_neighbors=5,
-        verbose=False,
         algorithm="auto",
         metric="euclidean",
         p=2,
         algo_params=None,
         metric_params=None,
         n_jobs=None,  # Ignored, here for sklearn API compatibility
+        verbose=False,
         output_type=None,
     ):
         super().__init__(verbose=verbose, output_type=output_type)

python/cuml/tests/test_sklearn_compatibility.py

@@ -118,24 +118,18 @@
     },
     RandomForestRegressor: {
         "check_estimator_tags_renamed": "No support for modern tags infrastructure",
-        "check_do_not_raise_errors_in_init_or_set_params": "RandomForestRegressor raises errors in init or set_params",
         "check_regressor_data_not_an_array": "RandomForestRegressor does not handle non-array data",
-        "check_dict_unchanged": "RandomForestRegressor modifies input dictionaries",
     },
     RandomForestClassifier: {
         "check_estimator_tags_renamed": "No support for modern tags infrastructure",
-        "check_do_not_raise_errors_in_init_or_set_params": "RandomForestClassifier raises errors in init or set_params",
         "check_classifier_data_not_an_array": "RandomForestClassifier does not handle non-array data",
-        "check_dict_unchanged": "RandomForestClassifier modifies input dictionaries",
     },
     KNeighborsClassifier: {
         "check_estimator_tags_renamed": "No support for modern tags infrastructure",
-        "check_do_not_raise_errors_in_init_or_set_params": "KNeighborsClassifier raises errors in init or set_params",
         "check_classifier_data_not_an_array": "KNeighborsClassifier does not handle non-array data",
     },
     KNeighborsRegressor: {
         "check_estimator_tags_renamed": "No support for modern tags infrastructure",
-        "check_do_not_raise_errors_in_init_or_set_params": "KNeighborsRegressor raises errors in init or set_params",
         "check_regressor_data_not_an_array": "KNeighborsRegressor does not handle non-array data",
         "check_supervised_y_2d": "KNeighborsRegressor does not handle 2D y",
     },
@@ -182,7 +176,7 @@
     },
     TSNE: {
         "check_estimator_tags_renamed": "No support for modern tags infrastructure",
-        "check_dont_overwrite_parameters": "TSNE overwrites parameters during fit",
+        "check_dont_overwrite_parameters": "TSNE only supports n_components = 2",
         "check_pipeline_consistency": "TSNE results are not deterministic",
         "check_methods_sample_order_invariance": "TSNE results depend on sample order",
         "check_methods_subset_invariance": "TSNE results depend on data subset",