diff --git a/paddle/fluid/operators/elementwise_mul_op.cc b/paddle/fluid/operators/elementwise_mul_op.cc
index ba343909bb87b4f2efa56c0a4ff664b278e90c60..7cd67e74de6b9c4fbc718f60b4f671ccab2f9956 100644
--- a/paddle/fluid/operators/elementwise_mul_op.cc
+++ b/paddle/fluid/operators/elementwise_mul_op.cc
@@ -15,7 +15,7 @@ limitations under the License. */
 #include "paddle/fluid/operators/elementwise_mul_op.h"
 #include "paddle/fluid/operators/elementwise_op.h"
 namespace ops = paddle::operators;
-REGISTER_ELEMWISE_OP(elementwise_mul, "Mul", "Out = X \\odot\\ Y");
+REGISTER_ELEMWISE_OP(elementwise_mul, "Mul", "Out = X \\\\odot Y");
 REGISTER_OP_CPU_KERNEL(
     elementwise_mul,
     ops::ElementwiseMulKernel<paddle::platform::CPUDeviceContext, float>,
diff --git a/python/paddle/fluid/layers/control_flow.py b/python/paddle/fluid/layers/control_flow.py
index 4fec6825078adf3e34d4b8f34aebf984346620b4..82f3d66ded9c12f6625cfe2cafeba956f5265efd 100644
--- a/python/paddle/fluid/layers/control_flow.py
+++ b/python/paddle/fluid/layers/control_flow.py
@@ -234,9 +234,56 @@ class BlockGuard(object):
 
 class ParallelDo(object):
     """
-    ParallelDo class.
+    ParallelDo is used to represent multi-thread data parallel processing.
 
-    ParallelDo class is used to create a ParallelDo.
+    Its vanilla implementation can be shown as the following (:math:`|` means
+    single thread and :math:`||||` means multiple threads)
+
+    .. code-block:: text
+
+      In the forward pass
+        |      Split input onto different devices
+        |      Copy parameter onto different devices
+        ||||   Compute forward pass in parallel
+        |      Merge output from different devices
+
+      In the backward pass
+        |      Split output@grad onto different devices
+        ||||   Compute backward pass in parallel
+        |      accumulate param@grad from different devices to the first device
+        |      Merge input@grad from different devices
+        |      Copy param@grad to the place of parallel_do_op
+
+    Examples:
+
+    .. code-block:: python
+
+      images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
+      label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+
+      # ParallelDo version & Single-thread version
+      if thread_num > 1:
+          places = fluid.layers.get_places(thread_num)
+          pd = fluid.layers.ParallelDo(places)
+          with pd.do():
+              images = pd.read_input(images)
+              label = pd.read_input(label)
+              predict = cnn_model(images)
+              cost = fluid.layers.cross_entropy(input=predict, label=label)
+
+              avg_cost = fluid.layers.mean(x=cost)
+              pd.write_output(avg_cost)
+
+          avg_cost = pd()
+          avg_cost = fluid.layers.mean(avg_cost)
+      else:
+          predict = cnn_model(images)
+          cost = fluid.layers.cross_entropy(input=predict, label=label)
+          avg_cost = fluid.layers.mean(x=cost)
+
+    .. warning::
+    
+       It will be soon deprecated, please use ParallelExecutor instead.
     """
 
     def __init__(self, places, use_nccl=False, name=None):
diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py
index 3816098383ff29450b53200c4bf6ef4f368a09cd..ce14452c27c1b0278d024ca3e750c17932a21688 100644
--- a/python/paddle/fluid/layers/nn.py
+++ b/python/paddle/fluid/layers/nn.py
@@ -1210,6 +1210,41 @@ def sequence_conv(input,
 
 
 def sequence_softmax(input, param_attr=None, bias_attr=None, use_cudnn=True):
+    """
+    This function computes the softmax activation among all time-steps for each
+    sequence. The dimension of each time-step should be 1. Thus, the shape of
+    input Tensor can be either :math:`[N, 1]` or :math:`[N]`, where :math:`N` 
+    is the sum of the length of all sequences.
+
+    For i-th sequence in a mini-batch:
+
+    .. math::
+
+        Out(X[lod[i]:lod[i+1]], :) = \\frac{\exp(X[lod[i]:lod[i+1], :])}{\sum(\exp(X[lod[i]:lod[i+1], :]))}
+
+    For example, for a mini-batch of 3 sequences with variable-length,
+    each containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7],
+    then softmax will be computed among :math:`X[0:2, :]`, :math:`X[2:5, :]`,
+    :math:`X[5:7, :]`, and :math:`N` turns out to be 7.
+
+    Args:
+        input (Variable): The input variable which is a LoDTensor.
+        bias_attr (ParamAttr|None): attributes for bias
+        param_attr (ParamAttr|None): attributes for parameter
+        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn \
+        library is installed. Default: True
+    
+    Returns:
+        Variable: output of sequence_softmax
+
+    Examples:
+
+        .. code-block:: python
+
+             x = fluid.layers.data(name='x', shape=[7, 1],
+                              dtype='float32', lod_level=1)
+             x_sequence_softmax = fluid.layers.sequence_softmax(input=x)
+    """
     helper = LayerHelper('sequence_softmax', **locals())
     dtype = helper.input_dtype()
     softmax_out = helper.create_tmp_variable(dtype)