* [cherry-pick] :Refine the error log about runtime batch and max_batch_size. #22535
test=release/1.7

* fix comments
test=release/1.7

cherry-pick #22615

* optimizer var exlcude fluid.data var. test=develop

* fix compute priority. test=develop

[cherry-pick] #22551

当一个模型中有多个fc_lstm子图的时候，且其中fc共用了同一个persistable的bias，此时不应该将bias节点删除，只将非persistable的节点去除即可。

* support  golang inference

* cherry-pick #22555
test=release/1.7, test=develop

* cherry-pick #22621
test=release/1.7, test=develop

pretty print for communicator flag

* add deprecated for distribute transpiler, will delete it after 2.0.0, test=develop

* enhance load program state; test=develop

* optimize commet; test=develop

* split unittests in data loader test, test=release/1.7

* split unittests to different files, test=release/1.7

* remove repeat unittest, test=release/1.7

Post_training_quantization support set quant 8/16 bits 

* 1. load model from memory

2. scale is no longer added when saving inference model

test=develop

* raise ci coverage, test=develop

* supports saving weights to memory. test=develop

* raise ci coverage, test=develop

* fix PADDLE_ENFORCE messages, test=develop

* Add support for dynamic_decode(while) training. test=develop

* Fix assign_op and tensor_array_read_write_op after solving conflict. test=develop

* Fix test_rnn_decode_api.py. test=develop

* Refine docs for apis in rnn.py. test=develop

* Adjust outputs of dynamic_decode. test=develop

* Remove the force_cpu update in assign_op. test=develop

* Remove the force_cpu update in assign_op. test=develop

* Make RNNCell.get_initial_states support batch_dim_idx argument. test=develop

* Rename _create_array_outof_while as _create_array_out_of_while in rnn.py.

test=release/1.7

* add thread barrier for the compiled program

* Introduce Ernie NLP

* Fix error regarding incorrect attr type

test=release/1.7
Co-authored-by: NWojciech Uss <wojciech.uss@intel.com>

* update external lite, test=develop

* switch WITH_TESTING to OFF, test=develop

This PR makes assign op support LoDTensorArray and enable the loop_vars in
while_loop to support tuple or list.

* Add weight quantization in post_training_quanzitaion (#22445)

* [cherry-pick]Support int16 for Tensor (#22423)

* add int16 support, test=develop, test=release/1.7
Co-authored-by: NLeo Chen <chenqiuliang@baidu.com>

add cp27-cp27m-gcc82 and cp27-cp27mu-gcc82 branch to support gcc8.2 compile for paddle, test=develop (#22504) (#22524)

[cherry-pick] #22509

支持不依赖nccl进行编译。

多卡下，如果没有打开WITH_NCCL开关编译，多卡不能通信，则只能选择一张卡使用

[cherry-pick] #22484

支持不依赖nccl进行编译。

多卡下，如果没有打开WITH_NCCL开关编译，则只能使用单卡

test=develop

* optimize interpolate op.
test=develop

test=release/1.7

* Add dequant scale squash

* Correct dequant-scale squash test

test=release/1.7

* R-language inference support

* change pslib url of release1.7 to a stable commit
* test=develop

* revised readme/readme_cn for paddle (#22219)

* these are the revised readme/readme_cn pull-request, test=develop

* this is a readme pull request, test = develop

* these are the revised readme/readme_cn pull request,test = develop

* revised version 1.6.2->1.6.3,test=release/1.7

Co-authored-by: NManjunath Bhat <manjunathbhat9920@gmail.com>

[cherry-pick] #22384 and #22371

22384增加了WITH_NCCL开关

22371修改了fluid依赖lite的commit id

* add GeneralRoleMaker which is for general usage
* test=develop

* Fix GEO-SGD init & send Bug (#22375)

* test=develop, fix geo Send & Init

* test=release/1.7,test=develop, cherry-pick 8f36c395

* fix copy table bug (#22432)

* fix copy table bug of lost some feasign
* test=develop

* add hdfs ls retry time and sleep time, fix save inference (#22433)

* add hdfs ls retry time and sleep time, fix save inference
* test=develop