* optimize lstmp and sample_logits op, test=develop

* update op_use_default_grad_op_maker.spec, test=develop

* delete useless file,test=develop

* append 0 to dim variable to avoid memory reusage, test=develop

* test=release/1.4, fix hsigmoid dereference nullptr

* test=release/1.4, refine condition

* test=release/1.4, refine comments

* cherry-pick 16763,test=release/1.4

* cherry-pick 16763, test=release/1.4

* fix some comments, include cosine_decay,l2_normalize,pixel_shuffle

* Add api.spec, test=develop

* update api.spec, test=develop

* add api.spec,test=develop

* test=develop

* test=develop

* fix conflict,test=develop

* Add Pixel shuffle OP (#15782)

* add pixel_shuffle op

* add pixel_shuffle op, test=develop

* rewrite code, test=develop

* delete useless comment, test=develop

* Refine pixel_shuffle_op and unit testing

* refine code,test=develop

* refine .cu,test=develop

* fix unittest,test=develop

* Fix unit testing
test=develop

* resolve conflict, test=develop

* fix test, test=develop

* fix API, test=develop

* fix test datatype bug,test=develop

* polish comments,test=develop

* add API,test=develop

* test=develop

* Add Pixel_Shuffle OP,test=develop

* support python3,test=develop

* add include memory to travis CI bug,test=develop

* cherry-pick 16763,15782 , test=release/1.4

* Add linear learning warmup method

This warmup lr can be combinated with other learning rate strategies.
For example:
            decayed_lr = fluid.layers.linear_lr_warmup(
                fluid.layers.piecewise_decay(boundaries, lr_steps),
                warmup_steps, start_lr, end_lr)

test=develop

test=develop

test=develop

* test=develop, fix space_to_depth_doc

* test=develop, refine indent

* test=develop, refine code

* test=develop, add api spec

test=develop

test=develop

* add layer norm to Layers, add transformer prepare encoding

* little change

* finish encoder part

* add decoder part

* finish model part

* add test case and part of data feed

* add transformer test

* add to_parameter, add remove in set_attr

* test=develop, fix pos encoding bug, create_parameter with stantard name

* test=develop, rm dropout test in imperative

* test=develop, fix cpu error

* test=develop, fix minize bug

* test=develop, fix one hot not stop gradient

* test=develop, fix one hot not stop gradient

* test=develop, refine parameter name

* test=develop, fix transformer test in imperative mode

* test=develop, fix transformer test in imperative mode

* test=develop, fix boost and mkl download error

* test=develop, fix boost and mkl download error

* test=develop, fix ci and refine code

* test=develop, fix ci and refine code

* Add fsp operator.
1 Add unitest.
2. Add python API.
3. Add layer test.

* Add quantization strategy.
1. Add API.
2. Add unitest.

* Add distillatoin strategy.

* Add unitest config file for quantization

* Fix Copyright
test=develop

* Fix setup.py

* Fix document of layers.py.
test=develop

* Fix unitest in python3.
test=develop

* Fix documents.
test=develop

* 1. refine fsp op by batched gemm
2. remove unused import
test=develop

* Fix test_dist_se_resnext.
1. disable test distillation.
2. reset framework.py
test=develop

* Enable unitest of distillation after fixing Block._clone_variable
test=develop

* Fix cdn issue.
test=develop

* Add range op.
test=develop

* Add more unitests.
test=develop

* Fix API.spec
test=develop

* Fix API.spec
test=develop

* Fix API.spec
test=develop

* Rewrite gradient ProtoMaker for affine_channel_op to remove the Output as the input.
* Add act in Python API to make the act can be in-place by layer_help.py

* wip allreduce in op

* wip

* wip

* wip

* wip adding test

* wip for conflict with mp mode

* fix tests test=develop

* fix cpu build test=develop

* fix travis clang format test=develop

* fix cpu build test=develop

* update api.spec test=develop

* delete comment test=develop

* fix cpplint test=develop

* fix test=develop

* follow comment test=develop

* add file test=develop

* fix build test=develop

* update test=develop

* to be compatible with sync_bn, and fix mp mode in develop test=develop

* Make step_input support custom lod level.
test=develop

* Fix API.spec
test=develop

* Fix API.spec.
test=develop

* Fix API.spec
test=develop

* Add default value in document of step_input.
test=develop

* Fix document.
test=develop

* Fix API.spec
test=develop

test=develop

test=develop

test=develop

test=develop

* Support Sync Batch Norm.
* Note, do not enable it in one device.

Usage:

build_strategy = fluid.BuildStrategy()
build_strategy.sync_batch_norm = True
binary = fluid.compiler.CompiledProgram(tp).with_data_parallel(
        loss_name=loss_mean.name,
        build_strategy=build_strategy)

test=develop

test=develop