diff --git a/src/mednet/config/data/montgomery/datamodule.py b/src/mednet/config/data/montgomery/datamodule.py
index edb2cd7137631c9de70c559c863f3d87a3cfc2ec..e245482718fd8ba3f84641f10000fe26cf2fc12b 100644
--- a/src/mednet/config/data/montgomery/datamodule.py
+++ b/src/mednet/config/data/montgomery/datamodule.py
@@ -48,7 +48,6 @@ class RawDataLoader(_BaseRawDataLoader):
             where to find the image to be loaded, and an integer, representing the
             sample label.
 
-
         Returns
         -------
             The sample representation.
diff --git a/src/mednet/engine/saliency/completeness.py b/src/mednet/engine/saliency/completeness.py
index c9a961f7fe6d52ec3c65526fe6347485078671ac..6d9cf7b25b27d64f26f9e97507e791788ca301ca 100644
--- a/src/mednet/engine/saliency/completeness.py
+++ b/src/mednet/engine/saliency/completeness.py
@@ -68,6 +68,7 @@ def _calculate_road_scores(
 
     Returns
     -------
+    tuple[float, float, float]
         A 3-tuple containing floating point numbers representing the
         most-relevant-first average score (``morf``), least-relevant-first
         average score (``lerf``) and the combined value (``(lerf-morf)/2``).
@@ -143,6 +144,15 @@ def _process_sample(
         A sequence of percentiles (percent x100) integer values indicating the
         proportion of pixels to perturb in the original image to calculate both
         MoRF and LeRF scores.
+
+    Returns
+    -------
+    list
+        A list containing the following items for a particular sample:
+        * The relative path to the sample.
+        * The label.
+        * An index to the specified target_class.
+        * The computed ROAD scores.
     """
 
     name: str = sample[1]["name"][0]
diff --git a/src/mednet/engine/saliency/generator.py b/src/mednet/engine/saliency/generator.py
index 72b53d5131580575787d119c12faed6f4999045d..6ce13812686206e18a200f2b99a1bf87cc4f5c5f 100644
--- a/src/mednet/engine/saliency/generator.py
+++ b/src/mednet/engine/saliency/generator.py
@@ -24,7 +24,19 @@ def _create_saliency_map_callable(
     target_layers: list[torch.nn.Module] | None,
     use_cuda: bool,
 ):
-    """Creates a class activation map (CAM) instance for a given model."""
+    """Creates a class activation map (CAM) instance for a given model.
+
+    Parameters
+    ----------
+    algo_type
+        The algorithm to use for saliency map estimation.
+    model
+        Neural network model (e.g. pasa).
+    target_layers
+        The target layers to compute CAM for.
+    use_cuda
+        Whether to use cuda or not.
+    """
 
     import pytorch_grad_cam