Removing the experiment script from add fire (#4373)

* Removing the experiment script

* Removing the script

Author: vincentpierre

experiment_torch.py

@@ -21,7 +21,7 @@
 )
 from mlagents.trainers.cli_utils import parser
 from mlagents_envs.environment import UnityEnvironment
-from mlagents.trainers.settings import RunOptions, TestingConfiguration
+from mlagents.trainers.settings import RunOptions
 
 from mlagents.trainers.training_status import GlobalTrainingStatus
 from mlagents_envs.base_env import BaseEnv
@@ -35,8 +35,6 @@
 )
 from mlagents_envs import logging_util
 
-from mlagents_envs.registry import default_registry
-
 logger = logging_util.get_logger(__name__)
 
 TRAINING_STATUS_FILE_NAME = "training_status.json"
@@ -198,27 +196,16 @@ def create_unity_environment(
     ) -> UnityEnvironment:
         # Make sure that each environment gets a different seed
         env_seed = seed + worker_id
-        if TestingConfiguration.env_name == "":
-            return UnityEnvironment(
-                file_name=env_path,
-                worker_id=worker_id,
-                seed=env_seed,
-                no_graphics=no_graphics,
-                base_port=start_port,
-                additional_args=env_args,
-                side_channels=side_channels,
-                log_folder=log_folder,
-            )
-        else:
-            return default_registry[TestingConfiguration.env_name].make(
-                seed=env_seed,
-                no_graphics=no_graphics,
-                base_port=start_port,
-                worker_id=worker_id,
-                additional_args=env_args,
-                side_channels=side_channels,
-                log_folder=log_folder,
-            )
+        return UnityEnvironment(
+            file_name=env_path,
+            worker_id=worker_id,
+            seed=env_seed,
+            no_graphics=no_graphics,
+            base_port=start_port,
+            additional_args=env_args,
+            side_channels=side_channels,
+            log_folder=log_folder,
+        )
 
     return create_unity_environment
 

ml-agents/mlagents/trainers/learn.py

@@ -8,7 +8,7 @@
 from mlagents_envs.base_env import DecisionSteps, BehaviorSpec
 from mlagents_envs.timers import timed
 
-from mlagents.trainers.settings import TrainerSettings, TestingConfiguration
+from mlagents.trainers.settings import TrainerSettings
 from mlagents.trainers.trajectory import SplitObservations
 from mlagents.trainers.torch.networks import (
     SharedActorCritic,
@@ -57,10 +57,7 @@ def __init__(
         )  # could be much simpler if TorchPolicy is nn.Module
         self.grads = None
 
-        if TestingConfiguration.device != "cpu":
-            torch.set_default_tensor_type(torch.cuda.FloatTensor)
-        else:
-            torch.set_default_tensor_type(torch.FloatTensor)
+        torch.set_default_tensor_type(torch.FloatTensor)
 
         reward_signal_configs = trainer_settings.reward_signals
         reward_signal_names = [key.value for key, _ in reward_signal_configs.items()]
@@ -83,7 +80,7 @@ def __init__(
             tanh_squash=tanh_squash,
         )
 
-        self.actor_critic.to(TestingConfiguration.device)
+        self.actor_critic.to("cpu")
 
     def split_decision_step(self, decision_requests):
         vec_vis_obs = SplitObservations.from_observations(decision_requests.obs)

ml-agents/mlagents/trainers/policy/torch_policy.py

@@ -15,12 +15,7 @@
 from mlagents.trainers.ppo.optimizer_tf import PPOOptimizer
 from mlagents.trainers.trajectory import Trajectory
 from mlagents.trainers.behavior_id_utils import BehaviorIdentifiers
-from mlagents.trainers.settings import (
-    TrainerSettings,
-    PPOSettings,
-    TestingConfiguration,
-    FrameworkType,
-)
+from mlagents.trainers.settings import TrainerSettings, PPOSettings, FrameworkType
 from mlagents.trainers.components.reward_signals import RewardSignal
 
 try:
@@ -64,8 +59,6 @@ def __init__(
             PPOSettings, self.trainer_settings.hyperparameters
         )
         self.seed = seed
-        if TestingConfiguration.max_steps > 0:
-            self.trainer_settings.max_steps = TestingConfiguration.max_steps
         self.policy: Policy = None  # type: ignore
 
     def _process_trajectory(self, trajectory: Trajectory) -> None:

ml-agents/mlagents/trainers/ppo/trainer.py

@@ -46,13 +46,6 @@ def defaultdict_to_dict(d: DefaultDict) -> Dict:
     return {key: cattr.unstructure(val) for key, val in d.items()}
 
 
-class TestingConfiguration:
-    use_torch = True
-    max_steps = 0
-    env_name = ""
-    device = "cpu"
-
-
 class SerializationSettings:
     convert_to_barracuda = True
     convert_to_onnx = True