* polish from concat to conv_shift op doc

* small fix

* small fix

* proj init.

* add unittest and implementation.

* Polish the documentation for uniform_random and top_k ops

* Polishing more operators

polish operator doc

* add acc layer
* memory log level change from 3 to 10
* use gaussian random to init conv parameters
* use initializer
* fix import
* batch_norm use helper to create persistable var
* refine code
* train only 2 batches for test
* use g_program and g_init_program
* use XavierInitializer to init fc parameter

* Add LoDRankTable

LoD Rank Table stores the `level` of `lod` which is ordered by sequence
length in descending order. It is useful when implement dynamic RNN and
is shared by dynamic RNN memory, dynamic RNN slice input and dynamic
RNN slice output operators.

* Add InferVarType

Check whether param name is manually set when input is a sequence in fc layer

Update faq

Polish activation operator documentation

Refine activation function pointer for LSTM operator.

* Add English version of Android cross-compiling document

* Add English version of Android cross-compiling document

* Follow comments from Yi-qun and Kavya

* init

* Fix bug

* rename test_filw

* refine test

fix build doc

* Fix bug in lookup_table_op & layers

* Missing Act in layers

* Should += in CPU

* Remove check in python

* Fix bug in sequence_conv_pool()

* Fix a bug in test_recommender_system.py

* Just skip test_evaluator

* Adding the Xavier Initializer
* Addressing code review feedback

Update IntelOptimizedPaddle.md

Support saving figure to file

add deploy script for website

* refine evaluator op types

* update

* follow comments

* update

* fix v2 mnist case

* fix v2 mnist case

* update

* update

* hack auc evaluator for dense vec

* follow comments

* Adding design doc for model average (now called parameter_average)

* Updating title

* Updating image tag

* Updating review comments

deconv cudnn forward passed

* "add net drawer for visualizing the graph"

* "fix "

* "add dep"

* Init commit

* Make executor use ProgramDescBind

* Change Attribute from BlockDesc to BlockDescBind

* Since we will get the program desc in RNN, just BlockDesc is not
  enough.

* Add DeviceContext to Executor API

* Rewrite RNN

* Pass Python

* AddBiasOp does not care num_flatten_dims

* Stash

* Fix MacOS Compile

* Pass RNN forward

* add python test

* refactor test

* Make compile pass

* add gradopmaker

* First draft done

* Polish code

* add grad op maker and grad infershape

* Polish code

* Fix backward.cc bug

* Fix infershape

* Rename function

* add backward test

* simplify recurrent test

* Update

* Pass unittest

* Add comments & refine test

* Add comments

* refactor test

* Complete Unittest

* fix StepScopes enforce

* Remove unused unittest

* no type error

* Update

* Make RNN Pass unittest