[RLlib] TorchLearner: Don't call no_sync (DDP/multi-GPU) on non-t…

…orch modules. (ray-project#50760)

rllib/BUILD

@@ -299,6 +299,16 @@ py_test(
     srcs = ["tuned_examples/appo/pendulum_appo.py"],
     args = ["--as-test", "--num-cpus=6", "--num-env-runners=4"]
 )
+# MultiAgentPong (multi-GPU smoke test)
+py_test(
+    name = "learning_tests_multi_agent_pong_appo_multi_gpu",
+    main = "tuned_examples/appo/multi_agent_pong_appo.py",
+    tags = ["team:rllib", "exclusive", "learning_tests", "learning_tests_discrete", "torch_only", "multi_gpu"],
+    size = "large",
+    srcs = ["tuned_examples/appo/multi_agent_pong_appo.py"],
+    args = ["--stop-iters=3", "--num-agents=2", "--num-learners=2", "--num-gpus-per-learner=1", "--num-aggregator-actors-per-learner=1"]
+)
+
 
 #@OldAPIStack
 py_test(

rllib/core/learner/torch/torch_learner.py

@@ -156,7 +156,9 @@ def _uncompiled_update(
         with contextlib.ExitStack() as stack:
             if self.config.num_learners > 1:
                 for mod in self.module.values():
-                    stack.enter_context(mod.no_sync())
+                    # Skip non-torch modules, b/c they may not have the `no_sync` API.
+                    if isinstance(mod, torch.nn.Module):
+                        stack.enter_context(mod.no_sync())
             postprocessed_gradients = self.postprocess_gradients(gradients)
             self.apply_gradients(postprocessed_gradients)
 

rllib/tuned_examples/appo/multi_agent_pong_appo.py

@@ -0,0 +1,104 @@
+import gymnasium as gym
+
+from ray.rllib.algorithms.appo import APPOConfig
+from ray.rllib.connectors.env_to_module.frame_stacking import FrameStackingEnvToModule
+from ray.rllib.connectors.learner.frame_stacking import FrameStackingLearner
+from ray.rllib.core.rl_module.default_model_config import DefaultModelConfig
+from ray.rllib.core.rl_module.multi_rl_module import MultiRLModuleSpec
+from ray.rllib.core.rl_module.rl_module import RLModuleSpec
+from ray.rllib.env.wrappers.atari_wrappers import wrap_atari_for_new_api_stack
+from ray.rllib.env.multi_agent_env import make_multi_agent
+from ray.rllib.examples.rl_modules.classes.random_rlm import RandomRLModule
+from ray.rllib.utils.test_utils import add_rllib_example_script_args
+
+parser = add_rllib_example_script_args(
+    default_reward=0.0,
+    default_timesteps=20000000,
+    default_iters=400,
+)
+parser.set_defaults(
+    enable_new_api_stack=True,
+    env="ale_py:ALE/Pong-v5",
+    num_agents=2,
+)
+args = parser.parse_args()
+
+
+def _make_env_to_module_connector(env):
+    return FrameStackingEnvToModule(num_frames=4, multi_agent=True)
+
+
+def _make_learner_connector(input_observation_space, input_action_space):
+    return FrameStackingLearner(num_frames=4, multi_agent=True)
+
+
+def _env_creator(cfg):
+    return wrap_atari_for_new_api_stack(
+        gym.make(args.env, **cfg, **{"render_mode": "rgb_array"}),
+        dim=64,
+        framestack=None,
+    )
+
+
+MultiAgentPong = make_multi_agent(_env_creator)
+
+
+config = (
+    APPOConfig()
+    .environment(
+        MultiAgentPong,
+        env_config={
+            "num_agents": args.num_agents,
+            # Make analogous to old v4 + NoFrameskip.
+            "frameskip": 1,
+            "full_action_space": False,
+            "repeat_action_probability": 0.0,
+        },
+        clip_rewards=True,
+    )
+    .env_runners(
+        env_to_module_connector=_make_env_to_module_connector,
+    )
+    .learners(
+        num_aggregator_actors_per_learner=2,
+    )
+    .training(
+        learner_connector=_make_learner_connector,
+        train_batch_size_per_learner=500,
+        target_network_update_freq=2,
+        lr=0.0005 * ((args.num_learners or 1) ** 0.5),
+        vf_loss_coeff=1.0,
+        entropy_coeff=[[0, 0.01], [3000000, 0.0]],  # <- crucial parameter to finetune
+        # Only update connector states and model weights every n training_step calls.
+        broadcast_interval=5,
+        # learner_queue_size=1,
+        circular_buffer_num_batches=4,
+        circular_buffer_iterations_per_batch=2,
+    )
+    .rl_module(
+        rl_module_spec=MultiRLModuleSpec(
+            rl_module_specs={
+                "main": RLModuleSpec(
+                    model_config=DefaultModelConfig(
+                        vf_share_layers=True,
+                        conv_filters=[(16, 4, 2), (32, 4, 2), (64, 4, 2), (128, 4, 2)],
+                        conv_activation="relu",
+                        head_fcnet_hiddens=[256],
+                    ),
+                ),
+                "random": RLModuleSpec(module_class=RandomRLModule),
+            },
+        ),
+    )
+    .multi_agent(
+        policies={"main", "random"},
+        policy_mapping_fn=lambda aid, eps, **kw: "main" if aid == 0 else "random",
+        policies_to_train=["main"],
+    )
+)
+
+
+if __name__ == "__main__":
+    from ray.rllib.utils.test_utils import run_rllib_example_script_experiment
+
+    run_rllib_example_script_experiment(config, args)