* OFA demo for ernie (#493)

* cherry pick

* fix bug when paddle upgrade (#606)

* add jpg

* fix when param name is not None

* fix nas (#553)

* fix when lambda_distill is None in ofa (#550)

* fix

* fix when lambda is None

* fix when the task is mnli (#540)

* fix

* fix when mnli

* Fix ofa (#536)

* fix

* fix embedding
Co-authored-by: NBai Yifan <me@ethanbai.com>
Co-authored-by: NBai Yifan <me@ethanbai.com>

* fix

* fix embedding
Co-authored-by: NBai Yifan <me@ethanbai.com>

* for bert

* support 2.0

* fix bugs for ernie

* migrate remained api in common module

* fix bug
Co-authored-by: NBai Yifan <me@ethanbai.com>

* remove sensitive_pruner module

* migrate prune module

* fix some bug for migrating the prune module to 2.0-rc

* update some 1.8 api

* change num_worker from 16 to 1

* migrate common and core module

* fix a api in rl controller

* fix dataloader

* Added dygraph quantization.

* Added save_quant_model function

* Overload quant_aware, make dynamic and static model use the same interface.

* Use singledispatch to overload the quant aware for dygraph quantization

* Add unnitest for dygraph quant_aware

* Use 2.0 api for supporting paddle 2.0

* Using pre-commit to modify the coding style.

* Add unittest for user inputted quant_config.

* Add singledispath into requirements.txt
Co-authored-by: NBai Yifan <me@ethanbai.com>

* remove sensitive_pruner module

* migrate prune module

* migrate quant code to paddle2.0

* fix quant unittest

* fix demo

* post_quant_static: remove the cache params, add optimize_model params, test=develop
Co-authored-by: NBai Yifan <me@ethanbai.com>

* add adaptive pact threshold matching

* update readme

* fix coverage

* fix coverage

* add cn times

* add en times

* align tag

* align type

* auto prune demo issue 382

* prune

* Update README.md

update default dataset

* ci ut

* code style
Co-authored-by: Ngmm <38800877+mmglove@users.noreply.github.com>

* add support of conditional block for pruning

* support pruning conditional block

* add notes for code

* add ofa

* fix

* update connect

* add uniitest

* add new argument for easy load prune program

* add argument make load pruned program more easier

* change save function arg

* fix test

* fix format

* fix pruned model save and load, fix bugs of pruning depthwise conv

* add unit test of pruned model save and load

* temp delete

* add unit test

* paddleslim/quant/quanter.py

* fix bug

* fix pruned model save and load, fix bugs of pruning depthwise conv

* add unit test of pruned model save and load

* temp delete

* add unit test

* add some unit tests

* complete autoprune unit test

* add sensitive unit tests

* fix some unit test

* fix auto_pruner.py

* fix pruned model save and load, fix bugs of pruning depthwise conv

* add unit test of pruned model save and load

* temp delete

* add unit test

* fix bug when prune the depthwise convolution layer

* fix bug when prune the depthwise convolution layer

* fix bug when prune depthwise convolution layer

* fix pruner when prune depthwise convolution layer

* remove print from unit test

* fix dml issue

* disable multiprocess reader

* add other model

* add dml unittest

* fix log_softmax

* make dml compatible with 1.8 and 2.0 version paddle

* fix coverage