diff --git a/04.word2vec/README.cn.md b/04.word2vec/README.cn.md
index d43a118fa701f4b29f13b173b542d00e0a3d982d..bf5d51bec9c1c5d4df82830c7e57fc7710ddbcb7 100644
--- a/04.word2vec/README.cn.md
+++ b/04.word2vec/README.cn.md
@@ -444,11 +444,11 @@ def infer(use_cuda, params_dirname=None):
         # 用来查询embedding表获取对应的词向量，因此其形状大小是[1]。
         # recursive_sequence_lengths设置的是基于长度的LoD，因此都应该设为[[1]]
         # 注意recursive_sequence_lengths是列表的列表
-        data1 = [[211]]  # 'among'
-        data2 = [[6]]  # 'a'
-        data3 = [[96]]  # 'group'
-        data4 = [[4]]  # 'of'
-        lod = [[1]]
+        data1 = numpy.asarray([[211]], dtype=numpy.int64)  # 'among'
+        data2 = numpy.asarray([[6]], dtype=numpy.int64)  # 'a'
+        data3 = numpy.asarray([[96]], dtype=numpy.int64)  # 'group'
+        data4 = numpy.asarray([[4]], dtype=numpy.int64)  # 'of'
+        lod = numpy.asarray([[1]], dtype=numpy.int64)
 
         first_word = fluid.create_lod_tensor(data1, lod, place)
         second_word = fluid.create_lod_tensor(data2, lod, place)
diff --git a/04.word2vec/README.md b/04.word2vec/README.md
index 2002e4798aa3069fc60964a8d743f1deb4dc7632..eb3e44a3cab784372227a17cd99c890f2adf1f3b 100644
--- a/04.word2vec/README.md
+++ b/04.word2vec/README.md
@@ -409,11 +409,11 @@ def infer(use_cuda, params_dirname=None):
         # Used to query the embedding table to get the corresponding word vector, so its shape size is [1].
         # recursive_sequence_lengths sets the length based on LoD, so it should all be set to [[1]]
         # Note that recursive_sequence_lengths is a list of lists
-        data1 = [[211]] # 'among'
-        data2 = [[6]] # 'a'
-        data3 = [[96]] # 'group'
-        data4 = [[4]] # 'of'
-        lod = [[1]]
+        data1 = numpy.asarray([[211]], dtype=numpy.int64)  # 'among'
+        data2 = numpy.asarray([[6]], dtype=numpy.int64)  # 'a'
+        data3 = numpy.asarray([[96]], dtype=numpy.int64)  # 'group'
+        data4 = numpy.asarray([[4]], dtype=numpy.int64)  # 'of'
+        lod = numpy.asarray([[1]], dtype=numpy.int64)
 
         first_word = fluid.create_lod_tensor(data1, lod, place)
         second_word = fluid.create_lod_tensor(data2, lod, place)
diff --git a/04.word2vec/index.cn.html b/04.word2vec/index.cn.html
index b7b727cebb6150c529bee3869488268b7f85fc97..73cbd2ee29d2c243a9dfe1c69930878f54ecb6c6 100644
--- a/04.word2vec/index.cn.html
+++ b/04.word2vec/index.cn.html
@@ -486,11 +486,11 @@ def infer(use_cuda, params_dirname=None):
         # 用来查询embedding表获取对应的词向量，因此其形状大小是[1]。
         # recursive_sequence_lengths设置的是基于长度的LoD，因此都应该设为[[1]]
         # 注意recursive_sequence_lengths是列表的列表
-        data1 = [[211]]  # 'among'
-        data2 = [[6]]  # 'a'
-        data3 = [[96]]  # 'group'
-        data4 = [[4]]  # 'of'
-        lod = [[1]]
+        data1 = numpy.asarray([[211]], dtype=numpy.int64)  # 'among'
+        data2 = numpy.asarray([[6]], dtype=numpy.int64)  # 'a'
+        data3 = numpy.asarray([[96]], dtype=numpy.int64)  # 'group'
+        data4 = numpy.asarray([[4]], dtype=numpy.int64)  # 'of'
+        lod = numpy.asarray([[1]], dtype=numpy.int64)
 
         first_word = fluid.create_lod_tensor(data1, lod, place)
         second_word = fluid.create_lod_tensor(data2, lod, place)
diff --git a/04.word2vec/index.html b/04.word2vec/index.html
index f67aeba9d25328734006374787e7898ebdadaa48..ab5e1f68d43cf154d19e24361e7cf0f73fc5914d 100644
--- a/04.word2vec/index.html
+++ b/04.word2vec/index.html
@@ -451,11 +451,11 @@ def infer(use_cuda, params_dirname=None):
         # Used to query the embedding table to get the corresponding word vector, so its shape size is [1].
         # recursive_sequence_lengths sets the length based on LoD, so it should all be set to [[1]]
         # Note that recursive_sequence_lengths is a list of lists
-        data1 = [[211]] # 'among'
-        data2 = [[6]] # 'a'
-        data3 = [[96]] # 'group'
-        data4 = [[4]] # 'of'
-        lod = [[1]]
+        data1 = numpy.asarray([[211]], dtype=numpy.int64)  # 'among'
+        data2 = numpy.asarray([[6]], dtype=numpy.int64)  # 'a'
+        data3 = numpy.asarray([[96]], dtype=numpy.int64)  # 'group'
+        data4 = numpy.asarray([[4]], dtype=numpy.int64)  # 'of'
+        lod = numpy.asarray([[1]], dtype=numpy.int64)
 
         first_word = fluid.create_lod_tensor(data1, lod, place)
         second_word = fluid.create_lod_tensor(data2, lod, place)
diff --git a/04.word2vec/train.py b/04.word2vec/train.py
index eebf62234ff408e24159e6bf6895ca5f85ef1a2d..2b3a97276d8fbd35b78ec480f2d1ea4496e0f62a 100644
--- a/04.word2vec/train.py
+++ b/04.word2vec/train.py
@@ -206,11 +206,11 @@ def infer(use_cuda, params_dirname=None):
         # meaning there is only one level of detail and there is only one sequence of
         # one word on this level.
         # Note that recursive_sequence_lengths should be a list of lists.
-        data1 = [[numpy.int64(211)]]  # 'among'
-        data2 = [[numpy.int64(6)]]  # 'a'
-        data3 = [[numpy.int64(96)]]  # 'group'
-        data4 = [[numpy.int64(4)]]  # 'of'
-        lod = [[numpy.int64(1)]]
+        data1 = numpy.asarray([[211]], dtype=numpy.int64)  # 'among'
+        data2 = numpy.asarray([[6]], dtype=numpy.int64)  # 'a'
+        data3 = numpy.asarray([[96]], dtype=numpy.int64)  # 'group'
+        data4 = numpy.asarray([[4]], dtype=numpy.int64)  # 'of'
+        lod = numpy.asarray([[1]], dtype=numpy.int64)
 
         first_word = fluid.create_lod_tensor(data1, lod, place)
         second_word = fluid.create_lod_tensor(data2, lod, place)