diff --git a/qlib/contrib/model/pytorch_alstm.py b/qlib/contrib/model/pytorch_alstm.py
index 40c2f8226..80393c3ec 100644
--- a/qlib/contrib/model/pytorch_alstm.py
+++ b/qlib/contrib/model/pytorch_alstm.py
@@ -113,6 +113,9 @@ class ALSTM(Model):
             )
         )
 
+        np.random.seed(self.seed)
+        torch.manual_seed(self.seed)
+        
         self.ALSTM_model = ALSTMModel(
             d_feat=self.d_feat,
             hidden_size=self.hidden_size,
diff --git a/qlib/contrib/model/pytorch_gru.py b/qlib/contrib/model/pytorch_gru.py
index 5daf4707e..4ef4e2553 100755
--- a/qlib/contrib/model/pytorch_gru.py
+++ b/qlib/contrib/model/pytorch_gru.py
@@ -113,6 +113,9 @@ class GRU(Model):
             )
         )
 
+        np.random.seed(self.seed)
+        torch.manual_seed(self.seed)
+
         self.gru_model = GRUModel(
             d_feat=self.d_feat,
             hidden_size=self.hidden_size,
diff --git a/qlib/contrib/model/pytorch_lstm.py b/qlib/contrib/model/pytorch_lstm.py
index eef1680ec..59df51e42 100755
--- a/qlib/contrib/model/pytorch_lstm.py
+++ b/qlib/contrib/model/pytorch_lstm.py
@@ -113,6 +113,9 @@ class LSTM(Model):
             )
         )
 
+        np.random.seed(self.seed)
+        torch.manual_seed(self.seed)
+
         self.lstm_model = LSTMModel(
             d_feat=self.d_feat,
             hidden_size=self.hidden_size,
diff --git a/qlib/contrib/model/pytorch_nn.py b/qlib/contrib/model/pytorch_nn.py
index d324e27aa..a60227ff0 100644
--- a/qlib/contrib/model/pytorch_nn.py
+++ b/qlib/contrib/model/pytorch_nn.py
@@ -61,6 +61,7 @@ class DNNModelPytorch(Model):
         optimizer="gd",
         loss="mse",
         GPU="0",
+        seed=0,
         **kwargs
     ):
         # Set logger.
@@ -80,6 +81,7 @@ class DNNModelPytorch(Model):
         self.loss_type = loss
         self.visible_GPU = GPU
         self.use_GPU = torch.cuda.is_available()
+        self.seed = seed
 
         self.logger.info(
             "DNN parameters setting:"
@@ -94,6 +96,7 @@ class DNNModelPytorch(Model):
             "\noptimizer : {}"
             "\nloss_type : {}"
             "\neval_steps : {}"
+            "\nseed : {}"
             "\nvisible_GPU : {}"
             "\nuse_GPU : {}".format(
                 layers,
@@ -107,11 +110,15 @@ class DNNModelPytorch(Model):
                 optimizer,
                 loss,
                 eval_steps,
+                seed,
                 GPU,
                 self.use_GPU,
             )
         )
 
+        np.random.seed(self.seed)
+        torch.manual_seed(self.seed)
+
         if loss not in {"mse", "binary"}:
             raise NotImplementedError("loss {} is not supported!".format(loss))
         self._scorer = mean_squared_error if loss == "mse" else roc_auc_score
diff --git a/qlib/contrib/model/pytorch_sfm.py b/qlib/contrib/model/pytorch_sfm.py
index 228c0aee5..d845f6245 100644
--- a/qlib/contrib/model/pytorch_sfm.py
+++ b/qlib/contrib/model/pytorch_sfm.py
@@ -282,6 +282,9 @@ class SFM(Model):
             )
         )
 
+        np.random.seed(self.seed)
+        torch.manual_seed(self.seed)
+
         self.sfm_model = SFM_Model(
             d_feat=self.d_feat,
             output_dim=self.output_dim,