diff --git a/a1/bc_agent.py b/a1/bc_agent.py
index e75ba2a..b22f9b9 100644
--- a/a1/bc_agent.py
+++ b/a1/bc_agent.py
@@ -140,5 +140,5 @@ def _compute_actor_loss(self, norm_obs, norm_expert_a):
         TODO 1.2: Implement code to calculate the loss for training the policy.
         '''
         # placeholder
-        loss = torch.zeros(1)
+        loss = torch.zeros(1, device=self._device)
         return loss
diff --git a/a2/cem_agent.py b/a2/cem_agent.py
index cb1ba00..8655a2d 100644
--- a/a2/cem_agent.py
+++ b/a2/cem_agent.py
@@ -150,8 +150,8 @@ def _eval_candidates(self, candidates):
         n = candidates.shape[0]
 
         # placeholder
-        rets = torch.zeros(n, device=self._device)
-        ep_lens = torch.zeros(n, device=self._device)
+        rets = torch.zeros(n, dtype=torch.float64, device=self._device)
+        ep_lens = torch.zeros(n, dtype=torch.float64, device=self._device)
 
         return rets, ep_lens
 
diff --git a/a2/pg_agent.py b/a2/pg_agent.py
index 4058eb6..1f52f93 100644
--- a/a2/pg_agent.py
+++ b/a2/pg_agent.py
@@ -218,7 +218,7 @@ def _calc_critic_loss(self, norm_obs, tar_val):
         '''
         
         # placeholder
-        loss = torch.zeros(1)
+        loss = torch.zeros(1, device=self._device)
         return loss
 
     def _calc_actor_loss(self, norm_obs, norm_a, adv):
@@ -229,5 +229,5 @@ def _calc_actor_loss(self, norm_obs, norm_a, adv):
         '''
         
         # placeholder
-        loss = torch.zeros(1)
+        loss = torch.zeros(1, device=self._device)
         return loss
\ No newline at end of file
diff --git a/a3/dqn_agent.py b/a3/dqn_agent.py
index aa1c6c0..a1b6d60 100644
--- a/a3/dqn_agent.py
+++ b/a3/dqn_agent.py
@@ -172,7 +172,7 @@ def _compute_q_loss(self, norm_obs, a, tar_vals):
         '''
         
         # placeholder
-        loss = torch.zeros(1)
+        loss = torch.zeros(1, device=self._device)
         
         return loss