Damping continuous reward monitors

sharif1093 · Sep 17, 2019 · dad9311 · dad9311
1 parent 67192c4
commit dad9311
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/digideep/environment/explorer.py b/digideep/environment/explorer.py
@@ -240,7 +240,8 @@ def step(self):
             self.state["hidden_state"] = extract_keywise(pre_transition["agents"], "hidden_state")
             self.state["masks"] = np.array([0.0 if done_ else 1.0 for done_ in dones], dtype=np.float32).reshape((-1,1))
 
-            monitor("/reward/"+self.params["mode"]+"/continuous", np.mean(rewards))
+            # NOTE: Uncomment if you find useful information in the continuous rewards ...
+            # monitor("/reward/"+self.params["mode"]+"/continuous", np.mean(rewards))
 
         with KeepTime("render"):
             if self.params["render"]: