fix dropout doc

docs/api/api_python/nn/mindspore.nn.Dropout.rst

@@ -5,7 +5,7 @@ mindspore.nn.Dropout
 
     随机丢弃层。
 
-    Dropout是一种正则化手段，该算子根据丢弃概率 :math:`1 - keep\_prob` 。在训练过程中随机将一些神经元输出设置为0，通过阻止神经元节点间的相关性来减少过拟合。在推理过程中，此层返回与 `x` 相同的Tensor。
+    Dropout是一种正则化手段，该算子根据丢弃概率 :math:`1 - keep\_prob`，在训练过程中随机将一些神经元输出设置为0，通过阻止神经元节点间的相关性来减少过拟合。在推理过程中，此层返回与 `x` 相同的Tensor。
 
     论文 `Dropout: A Simple Way to Prevent Neural Networks from Overfitting <http://www.cs.toronto.edu/~rsalakhu/papers/srivastava14a.pdf>`_ 中提出了该技术，并证明其能有效地减少过度拟合，防止神经元共适应。更多详细信息，请参见 `Improving neural networks by preventing co-adaptation of feature detectors <https://arxiv.org/pdf/1207.0580.pdf>`_ 。
 

docs/api/api_python/ops/mindspore.ops.func_dropout.rst

@@ -3,7 +3,7 @@ mindspore.ops.dropout
 
 .. py:function:: mindspore.ops.dropout(x, p=0.5, seed0=0, seed1=0)
 
-    在训练期间，以服从伯努利分布的概率 `p` 随机将输入Tensor的某些值归零，起到减少神经元相关性的作用，避免过拟合。
+    在训练期间，以服从伯努利分布的概率 `p` 随机将输入Tensor的某些值归零，起到减少神经元相关性的作用，避免过拟合。此概率与 `ops.dropout` 和 `nn.dropout` 中的含义相反。
 
     参数：
         - **x** (Tensor) - dropout的输入，任意维度的Tensor，其数据类型为float16或float32。
@@ -12,8 +12,8 @@ mindspore.ops.dropout
         - **seed1** (int，可选) - 全局的随机种子，和算子层的随机种子共同决定最终生成的随机数。默认值：0。
 
     返回：
-        - **output** (Tensor) - shape和数据类型与 `x` 相同。
-        - **mask** (Tensor) - shape与 `x` 相同。
+        - **output** (Tensor) - 归零后的Tensor，shape和数据类型与 `x` 相同。
+        - **mask** (Tensor) - 用于归零的掩码，内部会按位压缩与对齐。
 
     异常：
         - **TypeError** - `p` 不是float。

mindspore/python/mindspore/ops/function/nn_func.py

@@ -1063,7 +1063,8 @@ def dropout(x, p=0.5, seed0=0, seed1=0):
     """
     During training, randomly zeroes some of the elements of the input tensor
     with probability `p` from a Bernoulli distribution. It plays the role of
-    reducing neuron correlation and avoid overfitting.
+    reducing neuron correlation and avoid overfitting. The meaning of probability
+    here is opposite to that in `ops.dropout` and `nn.dropout`.
 
     Args:
         x (Tensor): The input of Dropout, a Tensor of any shape with data type of float16 or float32.
@@ -1073,8 +1074,8 @@ def dropout(x, p=0.5, seed0=0, seed1=0):
         seed1 (int, optional): seed1 value for random generating. Default: 0.
 
     Returns:
-        - **output** (Tensor) - With the same shape and data type as `x`.
-        - **mask** (Tensor) - With the same shape as `x`.
+        - **output** (Tensor) - Zeroed tensor, with the same shape and data type as `x`.
+        - **mask** (Tensor) - Mask for zeroing, bitwise compression and alignment are performed internally.
 
     Raises:
         TypeError: If `p` is not a float.
@@ -1088,8 +1089,8 @@ def dropout(x, p=0.5, seed0=0, seed1=0):
     Examples:
         >>> x = Tensor(((20, 16), (50, 50)), mindspore.float32)
         >>> output, mask = ops.dropout(x, p=0.5)
-        >>> print(output.shape)
-        (2, 2)
+        >>> print(output.shape, mask.shape, mask.dtype)
+        (2, 2) (16,) UInt8
     """
     keep_prob = 1 - p
     dropout_ = P.Dropout(keep_prob=keep_prob, Seed0=seed0, Seed1=seed1)

mindspore/python/mindspore/ops/operations/nn_ops.py

@@ -6940,11 +6940,11 @@ class Dropout(PrimitiveWithCheck):
         ``Ascend`` ``GPU`` ``CPU``
 
     Examples:
-        >>> dropout = ops.Dropout3D(keep_prob=0.5)
+        >>> dropout = ops.Dropout(keep_prob=0.5)
         >>> x = Tensor(np.ones([1, 2, 3, 4, 5]), mindspore.float32)
         >>> output, mask = dropout(x)
-        >>> print(output.shape)
-        (1, 2, 3, 4, 5)
+        >>> print(output.shape, mask.shape, mask.dtype)
+        (1, 2, 3, 4, 5) (16,) UInt8
     """
 
     @prim_attr_register