Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into complete_backward_doc

paddle/framework/backward.md

@@ -52,18 +52,32 @@ In our design, the network itself is also a kind of operator. So the operators c
 
 given a forward network, it generates the backward network. We only care about the Gradients—`OutputGradients`,`InputGradients`.
 
-1. bla bla bla (yuyang)
+1. Op 
+
+   when the input forward network is a Op, return its gradient Operator Immediately.
 
 2. NetOp 
 
-   when the input forward network is a NetOp, it need to call the sub NetOp/Operators backward function recursively and ensure them done. During the process, we need to collect the `OutputGradients` name.
+   when the input forward network is a NetOp, it need to call the sub NetOp/Operators backward function recursively. During the process, we need to collect the `OutputGradients` name according to forward NetOp.
+
+   **shared variable**. As illustrated in the pictures, two operator's `Output` `Gradient` will overwirte their shared input variable.  
+
+   <p align="center">
+   <img src="./images/duplicate_op.png" width="70%" ><br/>
+
+   1. shared variable in two operators. 
+
+   </p>
+
+   Share variable between operators or same input variable used in multiple operators lead to a duplicate gradient variable. As demo show above, we need to rename gradient name recursively, and add a generic add operator replace the overwirte links. 
+
+   <p align="center">
+   <img src="images/duplicate_op2.png" width="90%" ><br/>
 
-   We share variable in the same scope, as a result, duplicate operator `OutputGradients` will overwirte then duplicate variable.  
+   2. replace shared variable gradient with `Add` Operator
 
-   ![./images/duplicate_op]()
+   </p>
 
-    Share variable between operators or same input variable used in multiple operators lead to a duplicate gradient variable. As demo show above, we need to rename gradient name recursively, and add a generic add operator instead. 
 
-![./images/duplicate_op2]()
 
-​	Then collect the sub graph OutputGradients/InputGradients as the NetOp's and return it.
+​	Then collect the sub graph `OutputGradients`/`InputGradients` as the NetOp's and return it.

python/paddle/v2/framework/tests/mnist.py

@@ -181,7 +181,7 @@ images = data_layer(name='pixel', dims=[BATCH_SIZE, 784])
 labels = data_layer(name='label', dims=[BATCH_SIZE])
 fc1 = fc_layer(net=forward_net, input=images, size=100, act="sigmoid")
 fc2 = fc_layer(net=forward_net, input=fc1, size=100, act="sigmoid")
-predict = fc_layer(net=forward_net, input=fc2, size=100, act="softmax")
+predict = fc_layer(net=forward_net, input=fc2, size=10, act="softmax")
 cost = cross_entropy_layer(net=forward_net, input=predict, label=labels)
 
 init_net.complete_add_op(True)