🐛 fix bug

      all_y_task = []

      for task in range(n_tasks):

        y_task = y[:, task]

        # Handle continuous class probabilites of positive class for binary

        if len(np.unique(y_task)) > n_classes:

          # Handle continuous class probabilites of positive class for binary

          if n_classes > 2:

            raise ValueError(

                "Cannot handle continuous probabilities for multiclass problems."

          # Add a task dimension to concatenate on

          y_task = np.expand_dims(y_task, 1)

          all_y_task.append(y_task)

        # Handle binary labels

        else:

          # Handle binary labels

          # make y_hot of shape (N, n_classes)

          y_task = to_one_hot(y_task, n_classes=n_classes)

          # Add a task dimension to concatenate on

      else:

        n_tasks = self.n_tasks

    # check whether n_tasks is int or not

    # This is because `normalize_weight_shape` require int value.

    assert isinstance(n_tasks, int)

    y_true = normalize_labels_shape(

  return jaccard_score(y, y_pred)

def pixel_error(y: np.ndarray, y_pred: np.ndarray):

def pixel_error(y: np.ndarray, y_pred: np.ndarray) -> float:

  """An error metric in case y, y_pred are images.

  Defined as 1 - the maximal F-score of pixel similarity, or squared

  return mean_absolute_error(y_true, y_pred)

def bedroc_score(y_true: np.ndarray, y_pred: np.ndarray, alpha=20.0):

def bedroc_score(y_true: np.ndarray, y_pred: np.ndarray, alpha: float = 20.0):

  """Compute BEDROC metric.

  BEDROC metric implemented according to Truchon and Bayley that modifies