Removed duplicate references to the private methods

autoqild/bayes_search/bayes_search_cv.py

@@ -140,27 +140,6 @@ def __init__(
         self.logger = logging.getLogger(BayesSearchCV.__name__)
 
     def _step(self, search_space, optimizer, evaluate_candidates, n_points=1):
-        """Generate n_jobs parameters and evaluate them in parallel.
-
-        Parameters
-        ----------
-        search_space : dict
-            The search space for the hyperparameters.
-
-        optimizer : skopt.optimizer.Optimizer
-            The optimizer instance.
-
-        evaluate_candidates : callable
-            The function to evaluate the candidates.
-
-        n_points : int, default=1
-            Number of parameter settings to sample in parallel.
-
-        Returns
-        -------
-        optimizer : skopt.optimizer.Optimizer
-            The updated optimizer instance.
-        """
         # get parameter values to evaluate
         params = optimizer.ask(n_points=n_points)
 
@@ -183,13 +162,6 @@ def _step(self, search_space, optimizer, evaluate_candidates, n_points=1):
         return optimizer.tell(params, [-score for score in local_results])
 
     def _run_search(self, evaluate_candidates):
-        """Run the search for the best parameters.
-
-        Parameters
-        ----------
-        evaluate_candidates : callable
-            The function to evaluate the candidates.
-        """
         # check if space is a single dict, convert to list if so
         search_spaces = self.search_spaces
         if isinstance(search_spaces, dict):

autoqild/mi_estimators/mine_estimator.py

@@ -96,11 +96,6 @@ class MineMIEstimator(MIEstimatorBase):
     features and labels are complex or non-linear, as the aggregation process helps to smooth out inconsistencies
     across individual model predictions.
 
-    Private Methods
-    ---------------
-    __pytorch_tensor_dataset__:
-        Create PyTorch tensor datasets for the input features and target labels.
-
     Example
     -------
     >>> estimator = MineMIEstimator(n_classes=3, n_features=10)
@@ -142,7 +137,7 @@ def __init__(
         self.models = []
         self.n_models = 0
 
-    def __pytorch_tensor_dataset__(self, X, y, i=2):
+    def pytorch_tensor_dataset(self, X, y, i=2):
         """Create PyTorch tensor datasets for the input features and target
         labels.
 
@@ -241,7 +236,7 @@ def fit(self, X, y, epochs=100000, verbose=0, **kwd):
             sum_loss = 0
             for iter_ in tqdm(range(epochs), total=epochs, desc="iteration"):
                 stat_net.zero_grad()
-                xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, i=iter_)
+                xy, xy_tilde = self.pytorch_tensor_dataset(X, y, i=iter_)
                 preds_xy = stat_net(xy)
                 preds_xy_tilde = stat_net(xy_tilde)
                 train_div = get_mine_loss(preds_xy, preds_xy_tilde, metric=self.loss_function)
@@ -253,7 +248,7 @@ def fit(self, X, y, epochs=100000, verbose=0, **kwd):
                     with torch.no_grad():
                         mi_hats = []
                         for _ in range(MON_ITER):
-                            xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, i=iter_)
+                            xy, xy_tilde = self.pytorch_tensor_dataset(X, y, i=iter_)
                             preds_xy = stat_net(xy)
                             preds_xy_tilde = stat_net(xy_tilde)
                             eval_div = get_mine_loss(
@@ -378,7 +373,7 @@ def decision_function(self, X, verbose=0):
         for model in self.models:
             for n_class in range(self.n_classes):
                 y = np.zeros(X.shape[0]) + n_class
-                xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, i=0)
+                xy, xy_tilde = self.pytorch_tensor_dataset(X, y, i=0)
                 score = model(xy).detach().numpy()
                 if scores is None:
                     scores = score
@@ -418,7 +413,7 @@ def estimate_mi(self, X, y, verbose=0, MON_ITER=1000, **kwargs):
         for model in self.models:
             mi_hats = []
             for iter_ in range(MON_ITER):
-                xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, i=iter_)
+                xy, xy_tilde = self.pytorch_tensor_dataset(X, y, i=iter_)
                 preds_xy = model(xy)
                 preds_xy_tilde = model(xy_tilde)
                 eval_div = get_mine_loss(preds_xy, preds_xy_tilde, metric=self.loss_function)

autoqild/mi_estimators/mine_estimator_mse.py

@@ -87,11 +87,6 @@ class MineMIEstimatorMSE(MIEstimatorBase):
     mi_val : float
         The final estimated mutual information value.
 
-    Private Methods
-    ---------------
-    __pytorch_tensor_dataset__:
-        Create PyTorch tensor datasets for the input features and target labels.
-
     Notes
     -----
     This class is particularly suited for scenarios involving hyperparameter tuning where the goal is to identify
@@ -141,7 +136,7 @@ def __init__(
         self.final_loss = 0
         self.mi_val = 0
 
-    def __pytorch_tensor_dataset__(self, X, y, batch_size=64, i=2):
+    def pytorch_tensor_dataset(self, X, y, batch_size=64, i=2):
         """Create PyTorch tensor datasets for the input features and target
         labels.
 
@@ -238,7 +233,7 @@ def fit(self, X, y, epochs=10000, batch_size=128, verbose=0, **kwd):
         sum_loss = 0
         for iter_ in tqdm(range(epochs), total=epochs, desc="iteration"):
             self.stat_net.zero_grad()
-            xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, batch_size=batch_size, i=iter_)
+            xy, xy_tilde = self.pytorch_tensor_dataset(X, y, batch_size=batch_size, i=iter_)
             preds_xy = self.stat_net(xy)
             preds_xy_tilde = self.stat_net(xy_tilde)
             train_div = get_mine_loss(preds_xy, preds_xy_tilde, metric=self.loss_function)
@@ -250,7 +245,7 @@ def fit(self, X, y, epochs=10000, batch_size=128, verbose=0, **kwd):
                 with torch.no_grad():
                     mi_hats = []
                     for _ in range(MON_ITER):
-                        xy, xy_tilde = self.__pytorch_tensor_dataset__(
+                        xy, xy_tilde = self.pytorch_tensor_dataset(
                             X, y, batch_size=batch_size, i=iter_
                         )
                         preds_xy = self.stat_net(xy)
@@ -320,7 +315,7 @@ def score(self, X, y, sample_weight=None, verbose=0):
             The score of the model using the mean squared error between the original and permuted samples loss.
         """
         torch.no_grad()
-        xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, batch_size=X.shape[0], i=0)
+        xy, xy_tilde = self.pytorch_tensor_dataset(X, y, batch_size=X.shape[0], i=0)
         preds_xy = self.stat_net(xy).cpu().detach().numpy().flatten()
         preds_xy_tilde = self.stat_net(xy_tilde).cpu().detach().numpy().flatten()
         mse = mean_squared_error(preds_xy, preds_xy_tilde)
@@ -354,7 +349,7 @@ def decision_function(self, X, verbose=0):
         scores = None
         for n_class in range(self.n_classes):
             y = np.zeros(X.shape[0]) + n_class
-            xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, batch_size=X.shape[0], i=0)
+            xy, xy_tilde = self.pytorch_tensor_dataset(X, y, batch_size=X.shape[0], i=0)
             score = self.stat_net(xy).cpu().detach().numpy()
             if scores is None:
                 scores = score
@@ -389,7 +384,7 @@ def estimate_mi(self, X, y, verbose=0, MON_ITER=100, **kwargs):
         """
         mi_hats = []
         for iter_ in range(MON_ITER):
-            xy, xy_tilde = self.__pytorch_tensor_dataset__(X, y, batch_size=X.shape[0], i=iter_)
+            xy, xy_tilde = self.pytorch_tensor_dataset(X, y, batch_size=X.shape[0], i=iter_)
             preds_xy = self.stat_net(xy)
             preds_xy_tilde = self.stat_net(xy_tilde)
             eval_div = get_mine_loss(preds_xy, preds_xy_tilde, metric=self.loss_function)