#917 - Overhauled transaction and mean reward writing.

- Separated out learning outputs from evaluation outputs

src/primaite/agents/agent.py

@@ -64,7 +64,11 @@ class AgentSessionABC(ABC):
         "The session timestamp"
         self.session_path = _get_session_path(self.session_timestamp)
         "The Session path"
-        self.checkpoints_path = self.session_path / "checkpoints"
+        self.learning_path = self.session_path / "learning"
+        "The learning outputs path"
+        self.evaluation_path = self.session_path / "evaluation"
+        "The evaluation outputs path"
+        self.checkpoints_path = self.learning_path / "checkpoints"
         self.checkpoints_path.mkdir(parents=True, exist_ok=True)
         "The Session checkpoints path"
 
@@ -205,7 +209,6 @@ class HardCodedAgentSessionABC(AgentSessionABC):
         self._env: Primaite = Primaite(
             training_config_path=self._training_config_path,
             lay_down_config_path=self._lay_down_config_path,
-            transaction_list=[],
             session_path=self.session_path,
             timestamp_str=self.timestamp_str
         )

src/primaite/agents/rllib.py

@@ -21,7 +21,6 @@ def _env_creator(env_config):
     return Primaite(
         training_config_path=env_config["training_config_path"],
         lay_down_config_path=env_config["lay_down_config_path"],
-        transaction_list=env_config["transaction_list"],
         session_path=env_config["session_path"],
         timestamp_str=env_config["timestamp_str"]
     )
@@ -106,7 +105,6 @@ class RLlibAgent(AgentSessionABC):
             env_config=dict(
                 training_config_path=self._training_config_path,
                 lay_down_config_path=self._lay_down_config_path,
-                transaction_list=[],
                 session_path=self.session_path,
                 timestamp_str=self.timestamp_str
             )

src/primaite/agents/sb3.py

@@ -34,7 +34,7 @@ class SB3Agent(AgentSessionABC):
             _LOGGER.error(msg)
             raise ValueError(msg)
 
-        self._tensorboard_log_path = self.session_path / "tensorboard_logs"
+        self._tensorboard_log_path = self.learning_path / "tensorboard_logs"
         self._tensorboard_log_path.mkdir(parents=True, exist_ok=True)
         self._setup()
         _LOGGER.debug(
@@ -49,7 +49,6 @@ class SB3Agent(AgentSessionABC):
         self._env = Primaite(
             training_config_path=self._training_config_path,
             lay_down_config_path=self._lay_down_config_path,
-            transaction_list=[],
             session_path=self.session_path,
             timestamp_str=self.timestamp_str
         )
@@ -108,10 +107,13 @@ class SB3Agent(AgentSessionABC):
 
         if not episodes:
             episodes = self._training_config.num_episodes
-
-        _LOGGER.info(f"Beginning evaluation for {episodes} episodes @"
-                     f" {time_steps} time steps...")
-
+        self._env.set_as_eval()
+        if deterministic:
+            deterministic_str = "deterministic"
+        else:
+            deterministic_str = "non-deterministic"
+        _LOGGER.info(f"Beginning {deterministic_str} evaluation for "
+                     f"{episodes} episodes @ {time_steps} time steps...")
         for episode in range(episodes):
             obs = self._env.reset()
 
@@ -123,6 +125,7 @@ class SB3Agent(AgentSessionABC):
                 if isinstance(action, np.ndarray):
                     action = np.int64(action)
                 obs, rewards, done, info = self._env.step(action)
+        _LOGGER.info(f"Finished evaluation")
 
     @classmethod
     def load(self):