update biac reward

alr_envs/mujoco/ball_in_a_cup/ball_in_a_cup_reward.py

@@ -73,7 +73,6 @@ class BallInACupReward(alr_reward_fct.AlrReward):
             # cost = self._get_stage_wise_cost(ball_in_cup, min_dist, dist_final, dist_ctxt)
             cost = 2 * (0.5 * min_dist + 0.5 * dist_final + 0.1 * dist_ctxt)
             reward = np.exp(-1 * cost) - 1e-4 * action_cost
-            stop_sim = True
             success = dist_final < 0.05 and dist_ctxt < 0.05
         else:
             reward = - 1e-4 * action_cost

alr_envs/mujoco/ball_in_a_cup/utils.py

@@ -28,7 +28,7 @@ def make_env(rank, seed=0):
                                dt=env.dt,
                                weights_scale=0.1,
                                zero_start=True,
-                               zero_goal=False
+                               zero_goal=True
                                )
 
         env.seed(seed + rank)
@@ -37,7 +37,6 @@ def make_env(rank, seed=0):
     return _init
 
 
-
 def make_simple_env(rank, seed=0):
     """
     Utility function for multiprocessed env.