Merge branch 'develop' of https://github.com/PaddlePaddle/models into joe_upgrade_pretrain_lm_and_lm
1）升级language_model模型内接口到paddle 1.8版本
2）升级pretrain下bert和xlnet两个模块型内接口到paddle1.8版本
3）升级训练模型（language: 1个模型，pretrain_language_model: 2个模型）
接口涉及：
a) save/load
b) DataLoader
c) set_gradient_clip
d) embedding
e) one_hot
f) 部分io接口；

* add distributed training and gpu decoding and pre-processing of video

* Add document for DALI installation

* update lac and seq2seq to api_1.8

* update lac and seq2seq to api_1.8

* update lac and seq2seq to api_1.8

* test=develop update one_hot usage according to 1.8

* test=develop update dataloader api by 1.8

* test=develop fix image shape

* upgrade_dialogue_models

* fix_upgrade_models_comments

* fix init

* update

* fix

* fix bugs

* fix bugs

* add wide_deep

* fix code style

* fix code style

* fix some bugs

* fix filename

* add ncf

* add download data

* add download data

* fix dygraph in gru

* fix dy gru4rec

* add new cls link
* fix content
* fix cv readme
* fix moodels readme

* fix init

* update

* add fid_score for gan

* update readme

* fix Readme

* fix README

* fix README

* fix README

* fix README

* fix README

* fix README

* fix Readme

* fix bugs

* fix bugs

* add wide_deep

* fix code style

* fix code style

* fix some bugs

* fix filename

* fix bugs

* fix bugs

* add wide_deep

* fix code style

* fix code style

* fix bugs

* fix bugs

* add tdm_demo

* fix some bugs

* edit README.md

* add models

* add ESMM

* fix some bug

* fix models because the gradient clip strategy has been upgraded,test=develop

* fix models because the gradient clip strategy has been upgraded,test=develop

* refine code, add infer module

* update readme

* refine code

* fix save error of xdeepfm
* test=develop

修复了文档中与实际代码命名有区别的问题，增加了如何启动分布式训练的通用介绍