* add batch_norm_layer

* add img_conv_group layer and test

* add check to Tensor.type()

* forward can run

* with backward

* change label data time from int32 to int64

* refine code

* follow comment

Feature/multigpu

* Cast Operator

Cast input variable to other data type

* Fix compile error

* Add cast op

* Follow comments

* Accuracy does not support float/double, only support integers
* Polish error message when an operator does not support some device.

* add sparse support for sum op

* typo fix

* fix gpu build error

* fix unittest error

* typo fix

* infer var type and shape in op_test

* follow comments

* fix build error

* bypass some unittests depend on NetOp

* support sparse output for lookup table grad op

* refine codes

* fix gpu build error

* fix lookup table grad gpu kernel

* fix ci

* fix ci

* fix ci

* fix bug in lookup_table_grad op

* fix bug in test_word2vec

* register double kernel for some operators

* set is_sparse=True in test_word2vec

* fix lookup table grad op CUDA kernel bug

* disable test_modified_huber_loss_op temporarily

* disable test_lstm_unit_op temporarily

* Shrink Operator.h

* Fix CI compile

* Adding the increment op for global step

* Changing list to single op as per code review feedback

* Add `dump_to_file()` for ProgrameDescBind in pybind

* Update

* Add utility.py

* typo

* Fix bugs

* Move add_feed/fetch_components to untility.py

* Compelete dump

* Follow comments

* Change output of Prune() from inference to pointer

* Expose Prune() to Python

* Compelete save/load API of inference model

* Fix errors

* Debuging

* Compelete unit tests

* follow comments

Batch norm latest

* Add debug logs in scope, meta_cache and memory

* Add missing deps

Comment out `test_lstm_unit_op` test.

Add merge model scripts

Add pool2d cudnn op

* fix compile error

* remove unittest

* disable huber loss unittest

add logs and fix a python path bug

refine check macro

add Max strategy for sequence_pool op

Add huber loss operator.