* init softmax grad op

* add compute code

* export Backward to python

* update test ,export op.type to python

* update python test, fix compute bug

* update unit test

* use eigen

* optimize eigen code

* add gpu test

* register softmax_grad GPU kernel and fix test bug

* typo

* follow comments

Move pybind.cc/tensor_bind.h to paddle::framework

Add @EMPTY@ to default no_grad_vars in `Backward`

Fix CPU compile

error fix on the front page

update MKLDNN cmakes

Move paddle/framework/detail/tensor-inl.h into framework/tensor_impl.h

Add Gradient Operator for mean

Refine InferShape for recurrent_network_op

"add no gradient"

Add the check of cudnn version in cudnnBatchNorm

ignore im2col if not necessary in conv 1 * 1

Refine compute code in operators

Install GCC 4.8 in Dockerfile so to work with Eigen GPU code

Add ClipLayer

* export Backward to python

add RowL2NormLayer

Return Reference Instead Pointer to GetEigenDevice

add wget in docker production image

enable operator gpu unittest