* add instance norm

* refine switch doc, test=develop, test=document_preview

* add description and sample for member functions, test=develop

* Rewrite the content of load_persistables_cn.rst.

* refine the fill_constant doc test=develop

* fix according the standard test=develop

* fix fill_constant doc test=develop

* fix from standard test=develop

* refine api comments

* refine api comments

* update docs

* update docs

* Rewrite the content of the chain in Reader_cn.rst.

* refine While doc, test=develop

* add refine Switch api doc, test=develop

* refine While, delete Switch, test=develop, test=document_preview

* refine api docs of scale, selu, shape, split, soft_relu. test=develop, test=document_preview

* refine sample codes for api docs, test=develop

* add result comments for sample codes, test=develop

* refine api_cn, test=document_preview

* update cn doc for floor op

* refine the description of floor op

* fix reciprocal doc spec test=develop

* fix doc test=develop

* fix according reviewer advice test=develop

* fix according the review advice test=develop

* fix according the review advice test=develop

* Rewrite the content of load_params_cn.rst.

* use default_main_program instead of _main_program_.

* fix fill_constant_batch_like doc test=develop

* fix doc test=develop

* fix typo test=develop

* fix according reviewer advice test=develop

* fix according review advice test=develop

* Refine fsp_matrix/deformable_conv/fsp_matrix/polygon_box_transform api.

* fix argmin doc
test=document_preview

* test=document_preview

* fix
test=document_preview

* fix relu6 doc
test=document_preview

* fix threshold

* fix
test=document_preview

* fix relu doc
test=document_preview

* fix relu doc

* fix
test=document_preview

* update save_vars and load_vars, test=develop

* modify L1DecayRegularizer, L1Decay, L2DecayRegularizer and L2Decay, test=develop

* add black lines, test=develop

* Rewrite firstn and shuffle functions, test=develop

* improve seq_pool api cn doc test=document_preview

* refine sequence last and first pool doc test=document_preview

* refine sequence_pool dataType test=document_preview

* fix input dtype

* add dataType

* fix datatype

* update LoDTensor_cn.rst, test=document_preview

* update LoDTensor_cn.rst, test=document_preview

* update description of LoDTensor

* follow comments

* add name en

* refine style

* refine doc: pool2d/3d, adaptive_pool2d/3d, grid_sampler, kldiv_loss,
spectral_norm, temporal_shift. test=document_preview

* refine doc: exponential_decay, inverse_time_decay, natural_exp_decay. test=document_preview

* refine doc. test=document_preview

* refine doc for learning_rate_scheduler. test=document_preview

* refine OP. test=document_preview

* fix kldiv_loss. test=document_preview

* fix spectral_norm formula. test=document_preview

* Fix cn doc of some api:
1. affine_grid
2. generate_proposal_labels
3. generate_proposals
4. rpn_target_assign
5. warpctc

* add name

* refine name doc

* fix the reduce api test=develop

* fix typo test=develop

* fix according reviewer advice test=develop

* fix doc for PipyReader test=develop

* fix doc for PipyReader test=develop

* fix according reviewer advice test=develop

* add fluid.embedding cn doc test=document_preview

* fix example code test=document_preview

* refine ref link

* fix param True style

* add fluid.one_hot zh doc

* refine code style

* fix dtype description

* add one_hot_cn.rst in fluid_cn

* add default value

* support LoDTensor

* modify embedding doc

* refine param_attr

* rm warning

* fix data

* add padding data update doc

* add np.dtype

* improve embeding cn doc test=document_preview

* refine text format test=document_preview

* refine return type test=document_preview

* add data example test=document_preview

* refine warining and return type

* refine param_attr

* modify warning

* add warning

* modify x into input

* refine return doc

* add padding data update doc

* add np.dtype

* refine the equal api doc test=develop

* refine doc test=develop

* fix according the standard doc test=develop

* fix the doc test=develop

* review fix test=develop

* fix review test=develop

* fix review test=develop

* fix from standard test=develop

* add fix test=develop

* add data-parallel model doc

* Polish DyGraph doc

* skip PipeReader in IO

* skip PipeReader in IO
test=develop

test=document_preview fix the defination of nms_eta in api_cn of op retinanet_detection_output (#1289)

* add data loader doc, test=develop

* follow comments, test=develop

* follow comments again, test=develop