multi-agent-rlenv 3.3.2__py3-none-any.whl → 3.3.5__py3-none-any.whl

marlenv/__init__.py

@@ -62,7 +62,7 @@ print(env.extras_shape) # (1, )
 If you want to create a new environment, you can simply create a class that inherits from `MARLEnv`. If you want to create a wrapper around an existing `MARLEnv`, you probably want to subclass `RLEnvWrapper` which implements a default behaviour for every method.
 """
 
-__version__ = "3.3.2"
+__version__ = "3.3.5"
 
 from . import models
 from . import wrappers

marlenv/adapters/overcooked_adapter.py

@@ -22,7 +22,7 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
         self._oenv = oenv
         assert isinstance(oenv.mdp, OvercookedGridworld)
         self._mdp = oenv.mdp
-        self.visualizer = StateVisualizer()
+        self._visualizer = StateVisualizer()
         shape = tuple(int(s) for s in self._mdp.get_lossless_state_encoding_shape())
         shape = (shape[2], shape[0], shape[1])
         super().__init__(
@@ -53,19 +53,19 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
         return self.state.timestep
 
     def _state_data(self):
-        state = np.array(self._mdp.lossless_state_encoding(self.state))
+        state = np.array(self._mdp.lossless_state_encoding(self.state), dtype=np.float32)
         # Use axes (agents, channels, height, width) instead of (agents, height, width, channels)
         state = np.transpose(state, (0, 3, 1, 2))
         return state
 
     def get_state(self):
-        return State(self._state_data()[0], np.array([self.time_step / self.horizon]))
+        return State(self._state_data()[0], np.array([self.time_step / self.horizon], dtype=np.float32))
 
     def get_observation(self) -> Observation:
         return Observation(
             data=self._state_data(),
             available_actions=self.available_actions(),
-            extras=np.array([[self.time_step / self.horizon]] * self.n_agents),
+            extras=np.array([[self.time_step / self.horizon]] * self.n_agents, dtype=np.float32),
         )
 
     def available_actions(self):
@@ -88,6 +88,10 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
             info=info,
         )
 
+    def reset(self):
+        self._oenv.reset()
+        return self.get_observation(), self.get_state()
+
     def __deepcopy__(self, memo: dict):
         mdp = deepcopy(self._mdp)
         return Overcooked(OvercookedEnv.from_mdp(mdp, horizon=self.horizon))
@@ -96,6 +100,10 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
         return {"horizon": self.horizon, "mdp": self._mdp}
 
     def __setstate__(self, state: dict):
+        from overcooked_ai_py.mdp.overcooked_mdp import Recipe
+
+        mdp = state["mdp"]
+        Recipe.configure(mdp.recipe_config)
         self.__init__(OvercookedEnv.from_mdp(state["mdp"], horizon=state["horizon"]))
 
     def get_image(self):
@@ -107,7 +115,7 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
             ]:
                 rewards_dict[key] = value
 
-        image = self.visualizer.render_state(
+        image = self._visualizer.render_state(
             state=self._oenv.state,
             grid=self._mdp.terrain_mtx,
             hud_data=StateVisualizer.default_hud_data(self._oenv.state, **rewards_dict),

marlenv/models/env.py

@@ -127,7 +127,7 @@ class MARLEnv(ABC, Generic[ActionType, ActionSpaceType]):
 
     def seed(self, seed_value: int):
         """Set the environment seed"""
-        raise NotImplementedError("Method not implemented")
+        return
 
     @abstractmethod
     def get_observation(self) -> Observation:
@@ -158,9 +158,9 @@ class MARLEnv(ABC, Generic[ActionType, ActionSpaceType]):
         """Perform a random step in the environment."""
         return self.step(self.sample_action())
 
+    @abstractmethod
     def reset(self) -> tuple[Observation, State]:
         """Reset the environment and return the initial observation and state."""
-        return self.get_observation(), self.get_state()
 
     def render(self):
         """Render the environment in a window (or in console)"""

{multi_agent_rlenv-3.3.2.dist-info → multi_agent_rlenv-3.3.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: multi-agent-rlenv
-Version: 3.3.2
+Version: 3.3.5
 Summary: A strongly typed Multi-Agent Reinforcement Learning framework
 Project-URL: repository, https://github.com/yamoling/multi-agent-rlenv
 Author-email: Yannick Molinghen <yannick.molinghen@ulb.be>

{multi_agent_rlenv-3.3.2.dist-info → multi_agent_rlenv-3.3.5.dist-info}/RECORD

@@ -1,4 +1,4 @@
-marlenv/__init__.py,sha256=GFEcoE8jkA0vbuQyrVRpgsS-iLIelxjpwsB_6pGJGjs,3741
+marlenv/__init__.py,sha256=XbptMcX24_x5Tk4G6ff6hjI-FkE9MKHLULtCrmvEKl8,3741
 marlenv/env_builder.py,sha256=_rdwcWRqnHP7i4M4Oje1Y2nrEBKH9EzTpqOuw_PNUyw,5560
 marlenv/env_pool.py,sha256=R3WIrnQ5Zvff4HR1ecfkDmuO2zl7v1ywQ0K2_nvWFzs,1070
 marlenv/exceptions.py,sha256=gJUC_2rVAvOfK_ypVFc7Myh-pIfSU3To38VBVS_0rZA,1179
@@ -6,12 +6,12 @@ marlenv/mock_env.py,sha256=qB0fYFIfbopJf7Va8kCeVI5vsOy1-2JdEYe9gdV1Ruw,4761
 marlenv/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 marlenv/adapters/__init__.py,sha256=NEmuHPWz4SGQcgF7QuIeA0QaXK141JoYco-7mqj9Ghk,883
 marlenv/adapters/gym_adapter.py,sha256=Vx6ZrYI7kiNlJODmqyjXu9WCdbCr6trcMNot0pvYD74,2864
-marlenv/adapters/overcooked_adapter.py,sha256=Yf1xmjSgDOOSzR3QNqjxQ_ROFEdKUe_C8BF2nLlW3Us,6162
+marlenv/adapters/overcooked_adapter.py,sha256=B6umkRG6POUQNkeHgdIVuYRv8WjIA-3S1EVP6uaKwGY,6458
 marlenv/adapters/pettingzoo_adapter.py,sha256=9rwSc_b7qV3ChtEIevOkJvtIp7WoY3CVnu6L9DxlMB4,2852
 marlenv/adapters/pymarl_adapter.py,sha256=x__E90XpFbfSWhnBHtkcD6WYkmKki1LByNbUFoDBUcg,3416
 marlenv/adapters/smac_adapter.py,sha256=fOfKo1hL4ioKtM5qQGcwtfdkdwUEACjAZqaGmkoQUcU,8373
 marlenv/models/__init__.py,sha256=9M-rnj94nsdyO4zm_VEtyYBmde3iD2_eIY4bMB-IqCo,555
-marlenv/models/env.py,sha256=faezAKOIccBauOFeo9wu5sX32pFmP3AMmGyJzaTRJcM,7514
+marlenv/models/env.py,sha256=S0Yd75X4O7desCHrsdEdz045Aw1Zrtu-pkfz5obYQJI,7433
 marlenv/models/episode.py,sha256=ZGBx6lb2snrUhDgFEwHPV1dp-XvMA7k4quQVUNQxsP0,15140
 marlenv/models/observation.py,sha256=kAmh1hIoC2TGrZlGVzV0y4TXXCSrI7gcmG0raeoncYk,3153
 marlenv/models/spaces.py,sha256=pw8Sum_fHBkR-lyfTqUij4azMCNm8oBZrYZe4WVR7rA,7652
@@ -31,7 +31,7 @@ marlenv/wrappers/penalty_wrapper.py,sha256=v4_H8OEN2-yujLzRb6P7W7KwmXHtjAFsxcdp3
 marlenv/wrappers/rlenv_wrapper.py,sha256=C2XekgBIM4x3Wa2Mtsn7rihRD4ymC2hORI473Af0sfw,2962
 marlenv/wrappers/time_limit.py,sha256=CDIMMJPMyIDHSFxUJaC7nb7Kd86-07NgZeFhrpZm82o,3985
 marlenv/wrappers/video_recorder.py,sha256=d5AFu6qHqby9mOcBsYWYPxAPiK1vtnfMYdZ81AnCekI,2624
-multi_agent_rlenv-3.3.2.dist-info/METADATA,sha256=s9O2h4QdJZ4Ytq1hRiBFJPsAtWV0bd1JLxDx4MChaI0,4897
-multi_agent_rlenv-3.3.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-multi_agent_rlenv-3.3.2.dist-info/licenses/LICENSE,sha256=_eeiGVoIJ7kYt6l1zbIvSBQppTnw0mjnYk1lQ4FxEjE,1074
-multi_agent_rlenv-3.3.2.dist-info/RECORD,,
+multi_agent_rlenv-3.3.5.dist-info/METADATA,sha256=Yts2uzrt6q7VFqf6QNjwtYsuDlTHWDcQHKxKkFzptFg,4897
+multi_agent_rlenv-3.3.5.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+multi_agent_rlenv-3.3.5.dist-info/licenses/LICENSE,sha256=_eeiGVoIJ7kYt6l1zbIvSBQppTnw0mjnYk1lQ4FxEjE,1074
+multi_agent_rlenv-3.3.5.dist-info/RECORD,,