diff --git a/pina/callbacks/adaptive_refinment_callbacks.py b/pina/callbacks/adaptive_refinment_callbacks.py
index cb807ac..5af2cc8 100644
--- a/pina/callbacks/adaptive_refinment_callbacks.py
+++ b/pina/callbacks/adaptive_refinment_callbacks.py
@@ -95,7 +95,7 @@ class R3Refinement(Callback):
 
         # average loss
         avg = (tot_loss.mean()).to("cpu")
-        old_pts = {} # points to be retained
+        old_pts = {}  # points to be retained
         for location in self._sampling_locations:
             pts = trainer._model.problem.input_pts[location]
             labels = pts.labels
@@ -106,18 +106,18 @@ class R3Refinement(Callback):
                 pts = (pts[mask]).as_subclass(LabelTensor)
                 pts.labels = labels
                 old_pts[location] = pts
-                numb_pts =  self._const_pts[location] - len(old_pts[location])
+                numb_pts = self._const_pts[location] - len(old_pts[location])
                 # sample new points
                 trainer._model.problem.discretise_domain(
                     numb_pts, "random", locations=[location]
-                    )
+                )
 
-            else: # if no res greater than average, samples all uniformly
+            else:  # if no res greater than average, samples all uniformly
                 numb_pts = self._const_pts[location]
                 # sample new points
                 trainer._model.problem.discretise_domain(
                     numb_pts, "random", locations=[location]
-                    )
+                )
         # adding previous population points
         trainer._model.problem.add_points(old_pts)
 
@@ -154,8 +154,6 @@ class R3Refinement(Callback):
             const_pts[location] = len(pts)
         self._const_pts = const_pts
 
-
-
     def on_train_epoch_end(self, trainer, __):
         """
         Callback function called at the end of each training epoch.