* Simplize system_allocator and fix GPU_INFO

* Remove main_program/startup_program

* Fix optimizer

* Fix CI

* Follow comments

* "add DeviceContextPool"

* "add devicecontextpool in pybind"

* "add comments in python side "

* "fix static link error"

* "fix CI error"

* "add executor.py"

* "fix CI error"

* "add with gpu macro"

* "remove comment out codes"

* "add TODO items"

* "update init devices"

fix pipe reader on multi passes

* Make cast op support bool

Also add `elemwise_sub/mul/abs/clip` layers

* Make fuild.layers as a module

* Move layers as a module

* Split layers.py into layers module

* Fix CI

* Fix CI

Also add `elemwise_sub/mul/abs/clip` layers

Fix #6531

* "add registry interface"

* "move function to registry"

* "rename with meaningful name"

* "add exposed layers"

* "fixed based on comments"

* "remove unsed comments"

* Add fill_op

* Fix bug

* Fix documentation

* Address review comments

* "add global regularization"

* Polish `append_regularization_ops`

* init decoder_trainer

* can run

* fix lod

* add sharelod to cross_entropy_grad_op

* add avg_cost to fetch list

* modify learning rate

* can run

* optimie code

* add early exit

* fix print

* revert test_understand_sentiment_conv.py

* add act to fc

Add LRN efficient GPU implement

* init nmt

* encoder ready

* only generation implementation

waiting for dynamic rnn ready to train

* init python

* remove decoder temporary

* clean

* clean

* Add DataFeeder

A v2 API like data feeder for book demos.
We can feed data directly from reader.

* Fix CI

* Add an unittest for while/rnn op forward

* Add unittest for raw while op backward

* Fix CI

* Complete Dynamic RNN

* add crf_decoding layer

* fix some typo

* init trunk_evaluator

* add trunk_evaluator layer

* update chunk_eval_op and test, change int32 to int64

* fix a numeric problem

* change layers.trunk_evaluator to layers.trunk_eval

* fix typo

* add precision_val