update gym -> gymnasium, ray=2.0.0 -> ray=2.5.0

examples/gym/utils.py

@@ -16,7 +16,7 @@
 from typing import Any, Mapping
 
 import dm_env
-from gym import spaces
+from gymnasium import spaces
 import numpy as np
 import tree
 

examples/pettingzoo/sb3_train.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 """Binary to run Stable Baselines 3 agents on meltingpot substrates."""
 
-import gym
+import gymnasium as gym
 from meltingpot import substrate
 import stable_baselines3
 from stable_baselines3.common import callbacks

examples/pettingzoo/utils.py

@@ -15,7 +15,7 @@
 
 import functools
 
-from gym import utils as gym_utils
+from gymnasium import utils as gym_utils
 import matplotlib.pyplot as plt
 from meltingpot import substrate
 from ml_collections import config_dict
@@ -75,7 +75,7 @@ def reset(self, seed=None):
     timestep = self._env.reset()
     self.agents = self.possible_agents[:]
     self.num_cycles = 0
-    return utils.timestep_to_observations(timestep)
+    return utils.timestep_to_observations(timestep), {}
 
   def step(self, action):
     """See base class."""
@@ -92,7 +92,7 @@ def step(self, action):
       self.agents = []
 
     observations = utils.timestep_to_observations(timestep)
-    return observations, rewards, dones, infos
+    return observations, rewards, dones, dones, infos
 
   def close(self):
     """See base class."""

examples/requirements.txt

@@ -3,12 +3,11 @@ dm_env
 dm_meltingpot
 dm_tree
 dmlab2d
-gym
+gymnasium
 matplotlib
 ml_collections
-numpy<1.23  # Needed by Ray because it uses `np.bool`.
 pettingzoo>=1.22.3
-ray[rllib,default]==2.0.0
+ray[rllib,default]==2.5.0
 stable_baselines3
 supersuit>=3.7.2
 torch

examples/rllib/self_play_train.py

@@ -56,8 +56,7 @@ def get_config(
   # Gets the default training configuration
   config = ppo.PPOConfig()
   # Number of arenas.
-  # This is called num_rollout_workers in 2.2.0.
-  config.num_workers = num_rollout_workers
+  config.num_rollout_workers = num_rollout_workers
   # This is to match our unroll lengths.
   config.rollout_fragment_length = rollout_fragment_length
   # Total (time x batch) timesteps on the learning update.

examples/rllib/utils.py

@@ -17,7 +17,7 @@
 
 import dm_env
 import dmlab2d
-from gym import spaces
+from gymnasium import spaces
 from meltingpot import substrate
 from meltingpot.utils.policies import policy
 from ml_collections import config_dict
@@ -59,10 +59,10 @@ def __init__(self, env: dmlab2d.Environment):
         utils.spec_to_space(self._env.action_spec()))
     super().__init__()
 
-  def reset(self):
+  def reset(self, *args, **kwargs):
     """See base class."""
     timestep = self._env.reset()
-    return utils.timestep_to_observations(timestep)
+    return utils.timestep_to_observations(timestep), {}
 
   def step(self, action_dict):
     """See base class."""
@@ -76,7 +76,7 @@ def step(self, action_dict):
     info = {}
 
     observations = utils.timestep_to_observations(timestep)
-    return observations, rewards, done, info
+    return observations, rewards, done, done, info
 
   def close(self):
     """See base class."""
@@ -90,7 +90,7 @@ def get_dmlab2d_env(self):
   # which modes the `render` method supports.
   metadata = {'render.modes': ['rgb_array']}
 
-  def render(self, mode: str) -> np.ndarray:
+  def render(self) -> np.ndarray:
     """Render the environment.
 
     This allows you to set `record_env` in your training config, to record
@@ -109,10 +109,7 @@ def render(self, mode: str) -> np.ndarray:
     world_rgb = observation[0]['WORLD.RGB']
 
     # RGB mode is used for recording videos
-    if mode == 'rgb_array':
-      return world_rgb
-    else:
-      return super().render(mode=mode)
+    return world_rgb
 
   def _convert_spaces_tuple_to_dict(
       self,

examples/rllib/utils_test.py

@@ -14,7 +14,7 @@
 """Tests for utils.py."""
 
 from absl.testing import absltest
-from gym.spaces import discrete
+from gymnasium.spaces import discrete
 from meltingpot import substrate
 from meltingpot.configs.substrates import commons_harvest__open
 
@@ -43,7 +43,7 @@ def test_action_space_size(self):
 
   def test_reset_number_agents(self):
     """Test that reset() returns observations for all agents."""
-    obs = self._env.reset()
+    obs, _ = self._env.reset()
     self.assertLen(obs, self._num_players)
 
   def test_step(self):
@@ -56,7 +56,7 @@ def test_step(self):
       actions['player_' + str(player_idx)] = 1
 
     # Step
-    _, rewards, _, _ = self._env.step(actions)
+    _, rewards, _, _, _ = self._env.step(actions)
 
     # Check we have one reward per agent
     self.assertLen(rewards, self._num_players)
@@ -68,7 +68,7 @@ def test_render_modes_metadata(self):
   def test_render_rgb_array(self):
     """Test that render('rgb_array') returns the full world."""
     self._env.reset()
-    render = self._env.render('rgb_array')
+    render = self._env.render()
     self.assertEqual(render.shape, (144, 192, 3))