* fix softmax_with_cross_entropy cuda kernel overflow bug, test=develop

* replace old macro & for condition, test=develop

* polish details, test=develop

* add attr axis infershape. test=develop

* add CUDA kernel. test=develop

* fix unittest. test=develop

* fix unittest for soft_label. test=develop

* fix fp16 unittest. test=develop

* remove comment code. test=develop

* refine test for axis. test=develop

* add python api. test=develop

* fix doc. test=develop

* fix fp16 unittest. test=develop

* fix ngraph test. test=develop

* fix ENFORCE for test_imperative_transformer. test=develop

* fit for ngraph test. test=develop

* fix after rebase develop. test=develop

* fix doc. test=develop

* fix API.spec. test=develop

* fix test_layers. test=develop

* fix format. test=develop

* add use_cuda to inplace pass,test=develop

* add test softmax_with_xe_inplace test,test=develop

* speedup gc and inplace softmax_with_cross_entropy_grad
test=develop

* refine models gpu mem
Merge skip vars and warning messages of mem opt
remove relu mem opt
test=develop

* follow comments
test=develop

test=develop

test=develop

test=develop

test=develop

* clip softmax for fp16

* updates

* fuse xent support fp16 test=develop

* wip

* wip

* add simple row reduce

* wip fp16 accurate softmax

* add accurate softmax kernel for fp16 test=develop

* update test=develop

* fix cpu build test=develop

* update api.spec test=develop

* follow comments test=develop

* fix build test=develop

* fix trt build test=develop

* fix inference build test=develop

* fix merge test=develop

* update test=develop

* try fix build test=develop

* fix build test=develop

* rename real_exp test=develop

* fortest

* remove hacky kernels test=develop

* clean up test=develop

The macro should be defined by compiler rather than by source.

test=develop

* add ignore index

* update api.spec

* enhance softmax_with_cross_entropy

* Fix a critical bug in softmax_with_cross_entropy_op, which will lead to the wrong gradients.

* Enhance unit testing.

There are mainly following fixes:

- take `DeviceContext` as the template parameter of math functors and OpKernel instead of `Place`
- remove `eigen_device` interface in base class  `DeviceContext`
- remove `GetEigenDevice` interface in `ExecutionContext` and base class `DeviceContext`
- remove unused `platform::EigenDeviceConverter`
- rename `REGISTER_OP_GPU_KERNEL` to `REGISTER_OP_CUDA_KERNEL`
- rename `USE_GPU_ONLY_OP` to `USE_CUDA_ONLY_OP`

* Simplize Gradient Check

* Stash

* Extract apply_backward_pass to backward.py

Rename apply_backward_pass to append_backward_ops

* Use graph API to check gradient

* Fix ci

* Fix CI

* Fix backward for double precision

* Stash

* Fix CI

* Fix ci

* Ignore GRU test

* Ignore xe op

* Fix CI

* Fix softmax with xe gradient

The correct equation should be IG = OG * (d_softmax_with_xe())

* Fix typo

* Fix merge error

* Disable LRN

* Remove template parameter for Tensor methods

* Also check the type is correct when data()
* Simplize holder_

* Fix accuracy_op

* Register Code

* Change dataType to data_type

Follow PEP8

* Change name_convention to fit PEP8