ma setpiont update

domainlab/algos/trainers/fbopt_setpoint_ada.py

@@ -51,14 +51,12 @@ def transition_to(self, state):
         self.state_updater = state
         self.state_updater.accept(self)
 
-    def update_setpoint_ma(self, target):
+    def update_setpoint_ma(self, list_target):
         """
         using moving average
         """
-        temp_ma = self.coeff_ma * torch.tensor(target)
-        temp_ma += (1 - self.coeff_ma) * torch.tensor(self.setpoint4R)
-        temp_ma = temp_ma.tolist()
-        self.setpoint4R = temp_ma
+        target_ma = [self.coeff_ma * a + (1 - self.coeff_ma) *b for a, b in zip(self.setpoint4R, list_target)]
+        self.setpoint4R = target_ma
 
     def observe(self, epo_reg_loss, epo_task_loss):
         """
@@ -69,7 +67,7 @@ def observe(self, epo_reg_loss, epo_task_loss):
         self.state_task_loss = epo_task_loss
         if self.state_updater.update_setpoint():
             logger = Logger.get_logger(logger_name='main_out_logger', loglevel="INFO")
-            self.setpoint4R = self.state_epo_reg_loss
+            self.update_setpoint_ma(self.state_epo_reg_loss)
             logger.info(f"!!!!!set point updated to {self.setpoint4R}!")