diff --git a/pz_risk/wrappers/__init__.py b/pz_risk/wrappers/__init__.py
index e690d9a..7b53e06 100644
--- a/pz_risk/wrappers/__init__.py
+++ b/pz_risk/wrappers/__init__.py
@@ -1 +1,5 @@
 from .assert_invalid_actions import AssertInvalidActionsWrapper
+from .vector_observation import VectorObservationWrapper
+from .graph_observation import GraphObservationWrapper
+from .sparse_reward import SparseRewardWrapper
+from .dense_reward import DenseRewardWrapper
diff --git a/pz_risk/wrappers/assert_invalid_actions.py b/pz_risk/wrappers/assert_invalid_actions.py
index 23c8542..9b6daab 100644
--- a/pz_risk/wrappers/assert_invalid_actions.py
+++ b/pz_risk/wrappers/assert_invalid_actions.py
@@ -26,19 +26,21 @@ def step(self, action):
         elif state == GameState.Card:
             assert 0 <= action <= 1, 'Card Action should be 0 or 1: {}'.format(action)
         elif state == GameState.Attack:
-            edges = self.board.player_attack_edges(player)
-            assert action[0] <= 1, 'Attack Finished should be 0 or 1: {}'.format(action[0])
-            if action[0] == 0:
-                assert action[1] in edges, 'Attack Can not be performed from {} to {}'.format(gn(action[1][0]), gn(action[1][1]))
+            if not action[0]:
+                edges = self.board.player_attack_edges(player)
+                assert action[0] <= 1, 'Attack Finished should be 0 or 1: {}'.format(action[0])
+                if action[0] == 0:
+                    assert action[1] in edges, 'Attack Can not be performed from {} to {}'.format(gn(action[1][0]), gn(action[1][1]))
         elif state == GameState.Move:
             u = max(0, self.board.g.nodes[self.board.last_attack[1]]['units'] - 3)
             assert 0 <= action <= u, 'Move out of bound: {} ~ {}'.format(action, u)
         elif state == GameState.Fortify:
-            cc = self.board.player_connected_components(player)
-            c = [c for c in cc if action[1] in c][0]
-            assert 0 <= action[0] <= 1, 'Skip should be 0 or 1: {}'.format(action[0])
-            assert action[2] in c, 'Fortify Can not be performed from {} to {}'.format(gn(action[1]), gn(action[2]))
-            assert action[3] <= self.board.g.nodes[action[1]]['units'] ,'Fortify Can not be more than source units!'
+            if not action[0]:
+                cc = self.board.player_connected_components(player)
+                c = [c for c in cc if action[1] in c][0]
+                assert 0 <= action[0] <= 1, 'Skip should be 0 or 1: {}'.format(action[0])
+                assert action[2] in c, 'Fortify Can not be performed from {} to {}'.format(gn(action[1]), gn(action[2]))
+                assert action[3] <= self.board.g.nodes[action[1]]['units'] ,'Fortify Can not be more than source units!'
 
         super().step(action)