* add C-API inference document and provide demo in the doc

* fix Layer and Pool2D api, test=develop

* update label_smooth_cn.rst, test=develop

test=document_fix

Cherry-pick #1534

* Update save_load_variables.rst (#1520)

* update 1_6 release

* fix readme
test=document_preview

* add new tree index

* Update save_load_variables.rst (#1520)

* update 1_6 release

Fix the DOC of Windows compile , install and inference lib.

The cmake command for the Windows build installation should be modified.
Support download link for inference library of cuda10, update 1.5.2 download link for inference library.

* add new seq2seq api_cn, test=document_preview

* add seq2seq api_cn, test=document_preview

* add seq2seq api_cn, test=document_preview

* update compile_CentOS.md for compile in GPU, test=develop

* update compile_CentOS.md for compile in GPU, test=develop

* update compile_CentOS.md for compile in GPU, test=develop

* add stop_gradient notes in Dygraph, test=develop

* add stop_gradient notes in Dygraph, test=develop

* polish code in Dygraph, test=develop

* polish code in Dygraph, test=develop

* polish code in Dygraph, test=develop

* add Automatic pruning in Dygraph, test=develop

* imporve windows inference lib doc

* update sample images for c runtime library part of english version

modify fluid.layers.data to fluid.data and Synchronize Chinese doc and English doc

* fix Dygraph cn sample code, test=develop

* fix Dygraph doc format, test=develop

Fix fluid.layers.data of example code, Synchronize Chinese and English DOC of API:layers.sum.

* Remove benchmark from index.

* Refine the format of begin.
test=develop

* Refine the Chinese documentation of DynamicRNN.
test=develop
test=document_preview

* Refine the memory interface of DynamicRNN.
test=develop
test=document_preview

* Refine update_memory and output of DynamicRNN.
test=develop
test=document_preview

* Update the format of note and fix a error in example.
test=document_preview

* Add name and correct some examples.

* Refine the documentation according to the review comments.

* Add function list and the doc of function __call__.

* Add optional for x in step_input.
test=develop

* Update sub-titles.

* Change fluid.layers.data in examples to fluid.data.

* Add examples for step and static input.
test=develop

* Refine some sentences.
test=develop

* Add description of shape and data type for input and output variables.
test=develop

* fix apis, test=document_preview

* fix apis, test=document_preview

* fix apis, test=document_preview

* fix apis, test=document_preview

test=develop

new document to upload in this PR

the previous doc talks about how to run CTR training on Baidu Cloud K8S Cluster.

the new doc are modified in following points.

1.Run CTR Training on Baidu Cloud K8S + Huawei Volcano
2.Auto Split CTR model into sparse ones and dense ones.
3.Cube auto fetch Sparse model, Serving auto fetch dense model.
4.provide paddle serving client to test the ctr prediction.

* fix sequence_enumerate cn doc

* only surpport LoDTensor

* add input output shape

* test=document_fix

* fix space_to_depth cn doc

* change name's text

* add input output shape

* add code example

* test=document_fix

* revision

* change name's text

* add input output shape

* test=document_fix

* update piecewise decay, test=develop

* update, test=develop

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update exponential decay, test=develop

* update,test=develop

* update, test=develop

* update, test=develop

* update, test=develop

* update, test=document_fix

* update, test=document_fix

* set inverse time decay file to original, test=document_fix

* set piecewise decay to original, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix

* update, test=document_fix