diff --git a/lerobot/common/policies/sac/configuration_sac.py b/lerobot/common/policies/sac/configuration_sac.py
index 18ceee24..1d296bf1 100644
--- a/lerobot/common/policies/sac/configuration_sac.py
+++ b/lerobot/common/policies/sac/configuration_sac.py
@@ -95,6 +95,6 @@ class SACConfig:
             "use_tanh_squash": True,
             "log_std_min": -5,
             "log_std_max": 2,
-            "init_final": 0.01,
+            "init_final": 0.005,
         }
     )
diff --git a/lerobot/scripts/server/actor_server.py b/lerobot/scripts/server/actor_server.py
index fa4e34af..8284f024 100644
--- a/lerobot/scripts/server/actor_server.py
+++ b/lerobot/scripts/server/actor_server.py
@@ -163,10 +163,10 @@ def serve_actor_service(port=50052):
 
 def update_policy_parameters(policy: SACPolicy, parameters_queue: queue.Queue, device):
     if not parameters_queue.empty():
-        logging.debug("[ACTOR] Load new parameters from Learner.")
+        logging.info("[ACTOR] Load new parameters from Learner.")
         state_dict = parameters_queue.get()
         state_dict = move_state_dict_to_device(state_dict, device=device)
-        policy.load_state_dict(state_dict)
+        policy.load_state_dict(state_dict, strict=False)
 
 
 def act_with_policy(cfg: DictConfig, robot: Robot, reward_classifier: nn.Module):