chore: simplify post processing for classifiers

src/concrete/ml/sklearn/base.py

@@ -719,10 +719,6 @@ def n_classes_(self) -> Optional[numpy.ndarray]:  # pragma: no cover
 
         return len(self.classes_)
 
-    def _set_post_processing_params(self):
-        super()._set_post_processing_params()
-        self.post_processing_params.update({"classes_": self.classes_})
-
     def fit(self, X: Data, y: Target, **fit_parameters):
         X, y = check_X_y_and_assert_multi_output(X, y)
 
@@ -765,17 +761,16 @@ def predict(self, X: Data, fhe: Union[FheMode, str] = FheMode.DISABLE) -> numpy.
     def post_processing(self, y_preds: numpy.ndarray) -> numpy.ndarray:
         y_preds = super().post_processing(y_preds)
 
-        # Retrieve the number of target classes
-        classes = self.post_processing_params["classes_"]
+        # If the prediction array is 1D, which happens with some models such as XGBCLassifier or
+        # LogisticRegression models, we have a binary classification problem
+        n_classes = y_preds.shape[1] if y_preds.ndim > 1 and y_preds.shape[1] > 1 else 2
 
-        # If the predictions only has one dimension (i.e., binary classification problem), apply the
-        # sigmoid operator
-        if len(classes) == 2:
+        # For binary classification problem, apply the sigmoid operator
+        if n_classes == 2:
             y_preds = numpy_sigmoid(y_preds)[0]
 
-            # If the prediction array is 1D (which happens with some models such as XGBCLassifier
-            # models), transform the output into a 2D array [1-p, p], with p the initial
-            # output probabilities
+            # If the prediction array is 1D, transform the output into a 2D array [1-p, p],
+            # with p the initial output probabilities
             if y_preds.ndim == 1 or y_preds.shape[1] == 1:
                 y_preds = numpy.concatenate((1 - y_preds, y_preds), axis=1)