remove unused parameter scale_external_reward_by_intrinsic_reward_value

.gitignore

@@ -15,6 +15,7 @@ roboschool
 *.orig
 docs/site
 coach_env
+venv
 build
 rl_coach.egg*
 rl_coach_slim.egg*
@@ -32,4 +33,3 @@ trace_test*
 .cache/
 *.pyc
 coachenv
-

rl_coach/agents/agent.py

@@ -895,10 +895,7 @@ def observe(self, env_response: EnvResponse) -> bool:
             transition = self.update_transition_before_adding_to_replay_buffer(transition)
 
             # merge the intrinsic reward in
-            if self.ap.algorithm.scale_external_reward_by_intrinsic_reward_value:
-                transition.reward = transition.reward * (1 + self.last_action_info.action_intrinsic_reward)
-            else:
-                transition.reward = transition.reward + self.last_action_info.action_intrinsic_reward
+            transition.reward = transition.reward + self.last_action_info.action_intrinsic_reward
 
             # sum up the total shaped reward
             self.total_shaped_reward_in_current_episode += transition.reward
@@ -1026,7 +1023,7 @@ def emulate_observe_on_trainer(self, transition: Transition) -> bool:
         self.total_reward_in_current_episode += transition.reward
         self.shaped_reward.add_sample(transition.reward)
         self.reward.add_sample(transition.reward)
-        
+
         # create and store the transition
         if self.phase in [RunPhase.TRAIN, RunPhase.HEATUP]:
             # for episodic memories we keep the transitions in a local buffer until the episode is ended.

rl_coach/base_parameters.py

@@ -200,9 +200,6 @@ def __init__(self):
         # distributed agents params
         self.share_statistics_between_workers = True
 
-        # intrinsic reward
-        self.scale_external_reward_by_intrinsic_reward_value = False
-
         # n-step returns
         self.n_step = -1  # calculate the total return (no bootstrap, by default)