diff --git a/torchy_baselines/sac/sac.py b/torchy_baselines/sac/sac.py
index 56c741c..9d77cc2 100644
--- a/torchy_baselines/sac/sac.py
+++ b/torchy_baselines/sac/sac.py
@@ -251,7 +251,7 @@ class SAC(BaseRLModel):
             episode_num += n_episodes
             timesteps_since_eval += episode_timesteps
 
-            if self.num_timesteps > 0:
+            if self.num_timesteps > 0 and self.num_timesteps > self.learning_starts:
                 if self.verbose > 1:
                     print("Total T: {} Episode Num: {} Episode T: {} Reward: {}".format(
                         self.num_timesteps, episode_num, episode_timesteps, episode_reward))
diff --git a/torchy_baselines/td3/td3.py b/torchy_baselines/td3/td3.py
index 8fc00ae..7f34970 100644
--- a/torchy_baselines/td3/td3.py
+++ b/torchy_baselines/td3/td3.py
@@ -100,7 +100,7 @@ class TD3(BaseRLModel):
         observation = np.array(observation)
         with th.no_grad():
             observation = th.FloatTensor(observation.reshape(1, -1)).to(self.device)
-            return self.actor(observation).cpu().data.numpy()
+            return self.actor(observation).cpu().numpy()
 
     def predict(self, observation, state=None, mask=None, deterministic=True):
         """
@@ -222,7 +222,7 @@ class TD3(BaseRLModel):
             self.num_timesteps += episode_timesteps
             timesteps_since_eval += episode_timesteps
 
-            if self.num_timesteps > 0:
+            if self.num_timesteps > 0 and self.num_timesteps > self.learning_starts:
                 if self.verbose > 1:
                     print("Total T: {} Episode Num: {} Episode T: {} Reward: {}".format(
                         self.num_timesteps, episode_num, episode_timesteps, episode_reward))