PPO + JAX + EnvPool + Atari

.github/workflows/tests.yaml

@@ -193,7 +193,7 @@ jobs:
 
       # envpool tests
       - name: Install envpool dependencies
-        run: poetry install --with pytest,envpool
+        run: poetry install --with pytest,envpool,jax
       - name: Downgrade setuptools
         run: poetry run pip install setuptools==59.5.0
       - name: Run envpool tests

benchmark/ppo.sh

@@ -71,3 +71,21 @@ xvfb-run -a poetry run python -m cleanrl_utils.benchmark \
     --command "poetry run python cleanrl/ppo_continuous_action_isaacgym/ppo_continuous_action_isaacgym.py --track --capture-video --num-envs 8192 --num-steps 8 --update-epochs 5 --num-minibatches 4 --reward-scaler 0.01 --total-timesteps 600000000 --record-video-step-frequency 3660" \
     --num-seeds 3 \
     --workers 1
+
+
+poetry install --with envpool
+poetry run python -m cleanrl_utils.benchmark \
+    --env-ids Alien-v5 Amidar-v5 Assault-v5 Asterix-v5 Asteroids-v5 Atlantis-v5 BankHeist-v5 BattleZone-v5 BeamRider-v5 Berzerk-v5 Bowling-v5 Boxing-v5 Breakout-v5 Centipede-v5 ChopperCommand-v5 CrazyClimber-v5 Defender-v5 DemonAttack-v5 \
+    --command "poetry run python ppo_atari_envpool_xla_jax.py --track --wandb-project-name envpool-atari --wandb-entity openrlbenchmark" \
+    --num-seeds 3 \
+    --workers 1
+poetry run python -m cleanrl_utils.benchmark \
+    --env-ids DoubleDunk-v5 Enduro-v5 FishingDerby-v5 Freeway-v5 Frostbite-v5 Gopher-v5 Gravitar-v5 Hero-v5 IceHockey-v5 Jamesbond-v5 Kangaroo-v5 Krull-v5 KungFuMaster-v5 MontezumaRevenge-v5 MsPacman-v5 NameThisGame-v5 Phoenix-v5 Pitfall-v5 Pong-v5 \
+    --command "poetry run python ppo_atari_envpool_xla_jax.py --track --wandb-project-name envpool-atari --wandb-entity openrlbenchmark" \
+    --num-seeds 3 \
+    --workers 1
+poetry run python -m cleanrl_utils.benchmark \
+    --env-ids PrivateEye-v5 Qbert-v5 Riverraid-v5 RoadRunner-v5 Robotank-v5 Seaquest-v5 Skiing-v5 Solaris-v5 SpaceInvaders-v5 StarGunner-v5 Surround-v5 Tennis-v5 TimePilot-v5 Tutankham-v5 UpNDown-v5 Venture-v5 VideoPinball-v5 WizardOfWor-v5 YarsRevenge-v5 Zaxxon-v5 \
+    --command "poetry run python ppo_atari_envpool_xla_jax.py --track --wandb-project-name envpool-atari --wandb-entity openrlbenchmark" \
+    --num-seeds 3 \
+    --workers 1

cleanrl/ppo_atari_envpool.py

@@ -84,13 +84,6 @@ def __init__(self, env, deque_size=100):
         self.num_envs = getattr(env, "num_envs", 1)
         self.episode_returns = None
         self.episode_lengths = None
-        # get if the env has lives
-        self.has_lives = False
-        env.reset()
-        info = env.step(np.zeros(self.num_envs, dtype=int))[-1]
-        if info["lives"].sum() > 0:
-            self.has_lives = True
-            print("env has lives")
 
     def reset(self, **kwargs):
         observations = super().reset(**kwargs)
@@ -107,13 +100,8 @@ def step(self, action):
         self.episode_lengths += 1
         self.returned_episode_returns[:] = self.episode_returns
         self.returned_episode_lengths[:] = self.episode_lengths
-        all_lives_exhausted = infos["lives"] == 0
-        if self.has_lives:
-            self.episode_returns *= 1 - all_lives_exhausted
-            self.episode_lengths *= 1 - all_lives_exhausted
-        else:
-            self.episode_returns *= 1 - dones
-            self.episode_lengths *= 1 - dones
+        self.episode_returns *= 1 - infos["terminated"]
+        self.episode_lengths *= 1 - infos["terminated"]
         infos["r"] = self.returned_episode_returns
         infos["l"] = self.returned_episode_lengths
         return (