--- a
+++ b/docs/source/notes/encoders/mlp.rst
@@ -0,0 +1,34 @@
+MLP
+======================================================
+
+
+.. code-block:: python
+
+	class DeepPurpose.models.MLP(nn.Sequential)
+
+
+Multi-Layer Perceptron (MLP) (`Source <https://github.com/kexinhuang12345/DeepPurpose/blob/master/DeepPurpose/models.py#L218>`_) 
+is a class of feedforward artificial neural network. 
+An MLP consists of at least three layers of nodes: an input layer, a hidden layer and an output layer. 
+
+**constructor** create  an instance of MLP
+
+.. code-block:: python
+
+	__init__(self, input_dim, hidden_dim, hidden_dims)
+
+* **input_dim** (int) - dimension of input feature. 
+* **hidden_dim** (int) - dimension of hidden layer. 
+
+
+**Calling functions** implement the feedforward procedure of MLP. 
+
+
+.. code-block:: python
+
+	forward(self, v)
+
+* **v** (torch.Tensor) - input feature of MLP. 
+
+
+