Add classification_cost to v2 documentation

* refine the document

* Update README.md

refine io_convert and op_convert

* change to readme

* Update README.md

* Save the base of the new test.

* Create the notest_recognize_digits_conv and notest_recognize_digits_mlp

* precommit check

* Change the function name from _network to _program.
Update several variable names to make them more consistant.
Update the Inferencer construction call.

* Fix the incorrect format.

* Add label semantic examples with new api

* Address review comments

* Address review comment

* Fix clone function of Program to avoid memory leak.

* Fix inference_optimize function of framework.py.

* Reuse inference_optimize in framework.py.

* Add comments.

* fix trainer import error on ce

* fix setup.py.in

timeline for distributed training

Add comment to explain how to run inference test

fix fluid Metric

* init

* init

* add ut

* split singleton from base class

* add singleton

* ad singleton

Correct tensor copy misuse

Add the call of DropKids at the end of executor.Run to delete the local scopes created in operators (#10403)

* Add the call of DeleteScope to delete the memory of scope created by NewScope.

* Call DropKids at the end of executor.Run to delete all local scopes created in operators.

Feature/update sparse parameter

* adding operator to blas deps list

* use device_context instead to solve cycle deps

* "optimizer op support float64"

* "fix ci"

* "fix ftrl op"

* Load/save the params from the params_path

* Switch to use load_persistables and save_persistables

* Instaed of setup the executor to run program and scope. Pass the program to the load_persistables

Fluid new API: dist train without modifying code