* Add ernie link

* Add submodule

* Add README.md

* Update url of images.

* Add images.

* Add usage.md

* Add demo.md

* Fix demo.md

* Add model_zoo.md

* Fix pretrain models download in model zoo.

* Refine documents.
1. Rename python to Python
2. Fix README.md: vgg->VGG, resnet->ResNet, mobilenet->MobileNet
3. Rename flops to FLOPS

* Move docs of PaddleSlim from models.fluid.PaddleSlim to models.PaddleSlim

* 1. Introduce model format in README.md
2. Merge results of combined strategies into one table.

* Refine README.md

* Refine README
1. Fix some terms in README.md
2. Fix description of model format in README.md
3. Add Channel-Wise into feature list
4. Rename 'paddle mobile' to 'Paddle Mobile'
5. Add comments for dynamic quantization and static quantization
6. Add entry url for Model Zoo in README.md

* Refine README.

Add dygraph models

add elmo link to readme

test=develop

* Fix get_attention_feeder_for_infer (#2067)

* Fix distillation in slim demo. (#2107)

[cherry-pick]modified infer speed

* Fix eval.py in ocr ctc demo. (#1970)

* Refine readme in cycle gan. (#1972)

* 1. Refine readme
2. Fix some issues

* Refine README.md.

* Fix atention model. (#2059)

* 1. Enable GC strategy in run.sh (#2063)

2. Fix loading checkpoint in filter pruning strategy by adding init_model option into config file.

[cherry-pick] fix install_doc url for yolov3

mem_opt for yolov3.

update

[cherry-pick] update infer speed

[cherry-pick] fix yolo python3 compatible.

[cherry-pick] refine yolov3 param name

* fix bugs and add python3 supports
* minor changes

fix error when join thread in the end of data reading

Merge pull request #2037 from JesseyXujin/elmo_1

Elmo 1

gc flag for TSM.

[cherry-pick] polish readme for yolov3

fix - and _ mismatch, fix python3 bug in nonlocal codes