diff --git a/alr_envs/mujoco/ball_in_a_cup/utils.py b/alr_envs/mujoco/ball_in_a_cup/utils.py
index f889736..2e90404 100644
--- a/alr_envs/mujoco/ball_in_a_cup/utils.py
+++ b/alr_envs/mujoco/ball_in_a_cup/utils.py
@@ -88,13 +88,14 @@ def make_simple_env(rank, seed=0):
         env = DetPMPEnvWrapper(env,
                                num_dof=3,
                                num_basis=5,
-                               width=0.01,
+                               width=0.005,
+                               off=-0.1,
                                policy_type="motor",
                                start_pos=env.start_pos[1::2],
                                duration=3.5,
                                post_traj_time=4.5,
                                dt=env.dt,
-                               weights_scale=0.5,
+                               weights_scale=0.25,
                                zero_start=True,
                                zero_goal=True
                                )
diff --git a/alr_envs/utils/detpmp_env_wrapper.py b/alr_envs/utils/detpmp_env_wrapper.py
index f49862e..c667abf 100644
--- a/alr_envs/utils/detpmp_env_wrapper.py
+++ b/alr_envs/utils/detpmp_env_wrapper.py
@@ -10,6 +10,7 @@ class DetPMPEnvWrapper(gym.Wrapper):
                  num_dof,
                  num_basis,
                  width,
+                 off=0.01,
                  start_pos=None,
                  duration=1,
                  dt=0.01,
@@ -23,7 +24,7 @@ class DetPMPEnvWrapper(gym.Wrapper):
         self.num_dof = num_dof
         self.num_basis = num_basis
         self.dim = num_dof * num_basis
-        self.pmp = det_promp.DeterministicProMP(n_basis=num_basis, n_dof=num_dof, width=width, off=0.01,
+        self.pmp = det_promp.DeterministicProMP(n_basis=num_basis, n_dof=num_dof, width=width, off=off,
                                                 zero_start=zero_start, zero_goal=zero_goal)
         weights = np.zeros(shape=(num_basis, num_dof))
         self.pmp.set_weights(duration, weights)