diff --git a/examples/d3rlpy_training_api.py b/examples/d3rlpy_training_api.py
index 5d02607..deff62e 100644
--- a/examples/d3rlpy_training_api.py
+++ b/examples/d3rlpy_training_api.py
@@ -45,8 +45,8 @@ def __init__(self, estimator:MultiOutputClassifier) -> None:
         self.estimator = estimator
     
     def __call__(self, y:torch.Tensor, x:torch.Tensor):
-        x = x.numpy()
-        y = y.numpy()
+        x = x.cpu().numpy()
+        y = y.cpu().numpy()
         probs = self.estimator.predict_proba(X=x)
         res = []
         for i,out_prob in enumerate(probs):
diff --git a/examples/static.py b/examples/static.py
index b836721..42f40f1 100644
--- a/examples/static.py
+++ b/examples/static.py
@@ -40,8 +40,8 @@ def __init__(self, estimator:MultiOutputClassifier) -> None:
         self.estimator = estimator
     
     def __call__(self, y:torch.Tensor, x:torch.Tensor):
-        x = x.numpy()
-        y = y.numpy()
+        x = x.cpu().numpy()
+        y = y.cpu().numpy()
         probs = self.estimator.predict_proba(X=x)
         res = []
         for i,out_prob in enumerate(probs):
diff --git a/src/offline_rl_ope/OPEEstimators/DirectMethod.py b/src/offline_rl_ope/OPEEstimators/DirectMethod.py
index 98d193c..9495ae2 100644
--- a/src/offline_rl_ope/OPEEstimators/DirectMethod.py
+++ b/src/offline_rl_ope/OPEEstimators/DirectMethod.py
@@ -24,7 +24,7 @@ def __init__(self, model:QLearningAlgoBase) -> None:
     
     def get_q(self, state:torch.Tensor, action:torch.Tensor) -> torch.Tensor:
         values = torch.tensor(self.model.predict_value(
-            x=state.numpy(), action=action.numpy()))
+            x=state.cpu().numpy(), action=action.cpu().numpy()))
         return values
         
     def get_v(self, state:torch.Tensor) -> torch.Tensor:
diff --git a/src/offline_rl_ope/api/d3rlpy/Misc.py b/src/offline_rl_ope/api/d3rlpy/Misc.py
index 214db44..a48d616 100644
--- a/src/offline_rl_ope/api/d3rlpy/Misc.py
+++ b/src/offline_rl_ope/api/d3rlpy/Misc.py
@@ -9,5 +9,5 @@ def __init__(self, predict_func:D3rlpyAlgoPredictProtocal):
         self.predict_func = predict_func
         
     def __call__(self, x:torch.Tensor):
-        pred = self.predict_func(x.numpy())
+        pred = self.predict_func(x.cpu().numpy())
         return torch.Tensor(pred)
diff --git a/src/offline_rl_ope/api/d3rlpy/Scorers/IS.py b/src/offline_rl_ope/api/d3rlpy/Scorers/IS.py
index 48a0e69..656cb24 100644
--- a/src/offline_rl_ope/api/d3rlpy/Scorers/IS.py
+++ b/src/offline_rl_ope/api/d3rlpy/Scorers/IS.py
@@ -48,7 +48,7 @@ def __call__(
                            weights=self.cache[self.is_type].traj_is_weights, 
                            is_msk=self.cache.weight_msk, discount=self.discount
                            )
-        return res.numpy()
+        return res.cpu().numpy()