diff --git a/models/rank/pnn/config.yaml b/models/rank/pnn/config.yaml
index 0e1714a2a6139175dd5d9daed8db243a45dabe4d..b6304cd9e2fc9f6cc1179b575f6bc39681260a31 100755
--- a/models/rank/pnn/config.yaml
+++ b/models/rank/pnn/config.yaml
@@ -38,7 +38,6 @@ hyper_parameters:
         learning_rate: 0.0001
     sparse_feature_number: 1086460
     sparse_feature_dim: 9
-    is_sparse: False
     deep_input_size: 50
     use_inner_product: True
     num_field: 39
diff --git a/models/rank/pnn/model.py b/models/rank/pnn/model.py
index bc21cb01b5539c84c9d8d59cf36609498cc9d09b..8310cf712354cba5cdf8129e01d5a609135d049d 100755
--- a/models/rank/pnn/model.py
+++ b/models/rank/pnn/model.py
@@ -46,8 +46,6 @@ class Model(ModelBase):
             "hyper_parameters.sparse_feature_number", None)
         self.sparse_feature_dim = envs.get_global_env(
             "hyper_parameters.sparse_feature_dim", None)
-        self.is_sparse = envs.get_global_env("hyper_parameters.is_sparse",
-                                             False)
         self.deep_input_size = envs.get_global_env(
             "hyper_parameters.deep_input_size", 50)
         self.use_inner_product = envs.get_global_env(
@@ -75,7 +73,7 @@ class Model(ModelBase):
 
         first_weights_re = fluid.embedding(
             input=feat_idx,
-            is_sparse=self.is_sparse,
+            is_sparse=True,
             is_distributed=self.is_distributed,
             dtype='float32',
             size=[self.sparse_feature_number + 1, 1],
@@ -94,7 +92,7 @@ class Model(ModelBase):
 
         feat_embeddings_re = fluid.embedding(
             input=feat_idx,
-            is_sparse=self.is_sparse,
+            is_sparse=True,
             is_distributed=self.is_distributed,
             dtype='float32',
             size=[self.sparse_feature_number + 1, self.sparse_feature_dim],