diff --git a/BERT/convert_params.py b/BERT/convert_params.py
index 17cada954d4920dcb6a3c92d0371415655508a23..5cb95dbddfad200e402d94ccd835b5efb9dd2248 100644
--- a/BERT/convert_params.py
+++ b/BERT/convert_params.py
@@ -137,7 +137,7 @@ def parse(init_checkpoint):
         else:
             print("ignored param: %s" % var_name)
 
-        if fluid_param_name is not '':
+        if fluid_param_name != '':
             tf_fluid_param_name_map[var_name] = fluid_param_name
             tf_param_name_shape_map[var_name] = var_shape
             fluid_param_name = ''
diff --git a/BERT/model/bert.py b/BERT/model/bert.py
index ace3e609e64b90170f7144586189f155d3dd2220..e6348d54688057c101ceeb71083d286518798269 100644
--- a/BERT/model/bert.py
+++ b/BERT/model/bert.py
@@ -73,7 +73,7 @@ class BertModel(object):
         self._sent_emb_name = "sent_embedding"
         self._dtype = "float16" if use_fp16 else "float32"
 
-        # Initialize all weigths by truncated normal initializer, and all biases 
+        # Initialize all weigths by truncated normal initializer, and all biases
         # will be initialized by constant zero by default.
         self._param_initializer = fluid.initializer.TruncatedNormal(
             scale=config['initializer_range'])
@@ -109,7 +109,7 @@ class BertModel(object):
         emb_out = pre_process_layer(
             emb_out, 'nd', self._prepostprocess_dropout, name='pre_encoder')
 
-        if self._dtype is "float16":
+        if self._dtype == "float16":
             self_attn_mask = fluid.layers.cast(
                 x=self_attn_mask, dtype=self._dtype)
 
@@ -175,7 +175,7 @@ class BertModel(object):
                 name='mask_lm_trans_fc.w_0',
                 initializer=self._param_initializer),
             bias_attr=fluid.ParamAttr(name='mask_lm_trans_fc.b_0'))
-        # transform: layer norm 
+        # transform: layer norm
         mask_trans_feat = pre_process_layer(
             mask_trans_feat, 'n', name='mask_lm_trans')
 
diff --git a/BERT/run_squad.py b/BERT/run_squad.py
index efc0d9c73bc5c995c55b2aa93aabc8fc71155c99..1f7da456c5a25dde99bcb39d46b3ee15a339dd94 100644
--- a/BERT/run_squad.py
+++ b/BERT/run_squad.py
@@ -17,11 +17,13 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 
+import argparse
+import collections
+import multiprocessing
 import os
 import time
-import argparse
+
 import numpy as np
-import collections
 
 import paddle
 import paddle.fluid as fluid
diff --git a/ERNIE/batching.py b/ERNIE/batching.py
index beea08241c7e3607ec226895901bef59ca097237..797ab9f5938324df07ee3506870fc4cc21d6e75d 100644
--- a/ERNIE/batching.py
+++ b/ERNIE/batching.py
@@ -19,6 +19,8 @@ from __future__ import print_function
 
 import numpy as np
 
+from six.moves import xrange
+
 
 def mask(batch_tokens,
          seg_labels,
diff --git a/ERNIE/finetune/classifier.py b/ERNIE/finetune/classifier.py
index b5609af32f8399228e5b2450d07760e253ea7c60..37415fb97f9b3e425d60ff414fe548d83c5f13b1 100644
--- a/ERNIE/finetune/classifier.py
+++ b/ERNIE/finetune/classifier.py
@@ -22,6 +22,8 @@ import numpy as np
 
 import paddle.fluid as fluid
 
+from six.moves import xrange
+
 from model.ernie import ErnieModel
 
 
diff --git a/ERNIE/finetune/sequence_label.py b/ERNIE/finetune/sequence_label.py
index 327c9e56f0d2ae6f10718a88bce440b79b63dc18..dab6a58c85587381c27a53a94306d360fc405ac6 100644
--- a/ERNIE/finetune/sequence_label.py
+++ b/ERNIE/finetune/sequence_label.py
@@ -25,6 +25,8 @@ import multiprocessing
 import paddle
 import paddle.fluid as fluid
 
+from six.moves import xrange
+
 from model.ernie import ErnieModel
 
 def create_model(args,
diff --git a/ERNIE/model/ernie.py b/ERNIE/model/ernie.py
index 69d6e112eb97855e933800b57d5eeb560caf2257..e42b2f4558097f90ed1732dc0a7dd29b3a011aee 100644
--- a/ERNIE/model/ernie.py
+++ b/ERNIE/model/ernie.py
@@ -73,7 +73,7 @@ class ErnieModel(object):
         self._sent_emb_name = "sent_embedding"
         self._dtype = "float16" if use_fp16 else "float32"
 
-        # Initialize all weigths by truncated normal initializer, and all biases 
+        # Initialize all weigths by truncated normal initializer, and all biases
         # will be initialized by constant zero by default.
         self._param_initializer = fluid.initializer.TruncatedNormal(
             scale=config['initializer_range'])
@@ -109,7 +109,7 @@ class ErnieModel(object):
         emb_out = pre_process_layer(
             emb_out, 'nd', self._prepostprocess_dropout, name='pre_encoder')
 
-        if self._dtype is "float16":
+        if self._dtype == "float16":
             self_attn_mask = fluid.layers.cast(
                 x=self_attn_mask, dtype=self._dtype)
 
@@ -175,7 +175,7 @@ class ErnieModel(object):
                 name='mask_lm_trans_fc.w_0',
                 initializer=self._param_initializer),
             bias_attr=fluid.ParamAttr(name='mask_lm_trans_fc.b_0'))
-        # transform: layer norm 
+        # transform: layer norm
         mask_trans_feat = pre_process_layer(
             mask_trans_feat, 'n', name='mask_lm_trans')