Adjustments to env class name handling in ZooToGymMultiAgentAdapter

Author: levitation

aintelope/agents/sb3_base_agent.py

@@ -655,7 +655,7 @@ def train(self, num_total_steps):
             )  # need to resolve the conf before passing to subprocesses since OmegaConf resolvers do not seem to work well in subprocesses
 
             env_wrapper = MultiAgentZooToGymAdapterZooSide(
-                self.env, self.env_classname, self.cfg
+                self.env, self.cfg, self.env_classname
             )
             self.models, self.exceptions = env_wrapper.train(
                 num_total_steps=num_total_steps,

tests/environments/test_savanna_safetygrid_parallel.py

@@ -248,11 +248,14 @@ def sb3_gym_test_thread_entry_point(
     gpu_index,
     num_total_steps,
     model_constructor,
+    env_classname,
     agent_id,
     checkpoint_filename,
     cfg,
     observation_space,
     action_space,
+    *args,
+    **kwargs,
 ):
     env_wrapper = MultiAgentZooToGymAdapterGymSide(
         pipe, agent_id, checkpoint_filename, observation_space, action_space
@@ -291,7 +294,7 @@ def test_multiagent_zoo_to_gym_wrapper_scalarized_rewards(execution_number):
     env.seed(execution_number)
 
     env_wrapper = MultiAgentZooToGymAdapterZooSide(
-        env, cfg=None
+        env, cfg=None, env_classname=None
     )  # cfg is unused at sb3_gym_test_thread_entry_point() function
     _, exceptions = env_wrapper.train(
         num_total_steps=None,  # unused at sb3_gym_test_thread_entry_point() function

tests/environments/test_savanna_safetygrid_sequential.py

@@ -233,11 +233,14 @@ def sb3_gym_test_thread_entry_point(
     gpu_index,
     num_total_steps,
     model_constructor,
+    env_classname,
     agent_id,
     checkpoint_filename,
     cfg,
     observation_space,
     action_space,
+    *args,
+    **kwargs,
 ):
     env_wrapper = MultiAgentZooToGymAdapterGymSide(
         pipe, agent_id, checkpoint_filename, observation_space, action_space
@@ -280,7 +283,7 @@ def sb3_gym_test_thread_entry_point(
 #    env.seed(execution_number)
 
 #    env_wrapper = MultiAgentZooToGymAdapterZooSide(
-#        env, cfg=None
+#        env, cfg=None, env_classname=None
 #    )  # cfg is unused at sb3_gym_test_thread_entry_point() function
 #    _, exceptions = env_wrapper.train(
 #        num_total_steps=None,  # unused at sb3_gym_test_thread_entry_point() function