diff --git a/network.py b/network.py
index 00e405e..f299e50 100644
--- a/network.py
+++ b/network.py
@@ -312,7 +312,7 @@ class Network:
         # start = time.time()
 
         list_of_moves = []
-
+        test_list = []
         # Prepping of data
         for idx, board in enumerate(boards):
             all_board_moves = []
@@ -321,16 +321,41 @@ class Network:
                 for state in all_states:
                     state = np.array(self.board_trans_func(state, player*-1)[0])
                     all_board_moves.append(state)
+                    test_list.append(state)
             list_of_moves.append(np.array(all_board_moves))
 
 
-        # print(time.time() - start)
-        # start = time.time()
+        list_of_lengths = [len(board) for board in list_of_moves]
 
-        # Running data through networks
+        start = time.time()
+        for i in range(len(test_list)):
+            self.model.predict_on_batch(np.array([state]))
+        print("Indiviual rolls:", time.time() - start)
         all_scores = [self.model.predict_on_batch(board) for board in list_of_moves]
+
+
+        start = time.time()
+        all_scores_legit = self.model.predict_on_batch(np.array(test_list))
+
+        split_scores = []
+        from_idx = 0
+        for length in list_of_lengths:
+            split_scores.append(all_scores_legit[from_idx:from_idx+length])
+            from_idx += length
+
+        transformed_splits = [tf.reduce_mean(scores) for scores in split_scores]
+
+        print(transformed_splits)
+
+
+
+
+        print("All in one:", time.time() - start)
+
         scores_means = [tf.reduce_mean(score) for score in all_scores]
 
+        print(scores_means)
+
         transformed_means = [x if player == 1 else (1-x) for x in scores_means]
 
         # print(time.time() - start)