diff --git a/python/paddle/base/data_feed_desc.py b/python/paddle/base/data_feed_desc.py
index 933d4dd1c66198..c4367c880c80d2 100644
--- a/python/paddle/base/data_feed_desc.py
+++ b/python/paddle/base/data_feed_desc.py
@@ -93,7 +93,7 @@ def __init__(self, proto_file):
 
     def set_batch_size(self, batch_size):
         r"""
-        Set :attr:`batch_size` in :ref:`api_base_DataFeedDesc` . :attr:`batch_size` can be changed during training.
+        Set :attr:`batch_size` in ``paddle.base.DataFeedDesc`` . :attr:`batch_size` can be changed during training.
 
         Examples:
             .. code-block:: python
diff --git a/python/paddle/base/framework.py b/python/paddle/base/framework.py
index 3a8e741d91c446..920eaa7895bd54 100644
--- a/python/paddle/base/framework.py
+++ b/python/paddle/base/framework.py
@@ -2112,7 +2112,7 @@ def dtype(self):
     @property
     def lod_level(self):
         """
-        Indicating ``LoD`` info of current Variable, please refer to  :ref:`api_base_LoDTensor_en` to check the meaning
+        Indicating ``LoD`` info of current Variable, please refer to  :ref:`api_paddle_Tensor` to check the meaning
         of ``LoD``
 
         **Notes**:
diff --git a/python/paddle/distributed/fleet/layers/mpu/mp_layers.py b/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
index d7febc350a5b52..cfc42e58acc1de 100644
--- a/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
+++ b/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
@@ -52,7 +52,7 @@ class VocabParallelEmbedding(paddle.nn.Layer):
         num_embeddings(int): One element which indicate the size of the dictionary of embeddings.
         embedding_dim(int): One element which indicate the size of each embedding vector respectively.
         weight_attr(ParamAttr|None): To specify the weight parameter property. Default: None, which means the
-            default weight parameter property is used. See usage for details in :ref:`api_ParamAttr` . In addition,
+            default weight parameter property is used. See usage for details in :ref:`api_paddle_ParamAttr` . In addition,
             user-defined or pre-trained word vectors can be loaded with the :attr:`param_attr` parameter.
             The local word vector needs to be transformed into numpy format, and the shape of local word
             vector should be consistent with :attr:`num_embeddings` . Then :ref:`api_paddle_nn_initializer_Assign`
diff --git a/python/paddle/nn/layer/common.py b/python/paddle/nn/layer/common.py
index e041ed505b2608..23b6381e316e56 100644
--- a/python/paddle/nn/layer/common.py
+++ b/python/paddle/nn/layer/common.py
@@ -1401,7 +1401,7 @@ class Embedding(Layer):
             such as :ref:`api_paddle_optimizer_adadelta_Adadelta` , :ref:`api_paddle_optimizer_adamax_Adamax` , :ref:`api_paddle_optimizer_lamb_Lamb`.
             In these case, sparse must be False. Default: False.
         weight_attr(ParamAttr, optional): To specify the weight parameter property. Default: None, which means the
-            default weight parameter property is used. See usage for details in :ref:`api_ParamAttr` . In addition,
+            default weight parameter property is used. See usage for details in :ref:`api_paddle_ParamAttr` . In addition,
             user-defined or pre-trained word vectors can be loaded with the :attr:`param_attr` parameter.
             The local word vector needs to be transformed into numpy format, and the shape of local word
             vector should be consistent with :attr:`num_embeddings` . Then :ref:`api_paddle_nn_initializer_Assign`