test=develop

test=develop

* "fix compile error"

* "fix ci"

* rerun ci
test=develop

* test=develop

rerun ci

This reverts commit 21bb9e91, reversing
changes made to 3fa68dc1.

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

* refine reduce by cub
* optimize KernelDepthwiseConvFilterGrad
* optimize depthwise conv and reduce mean and reduce sum
* fix bug: dilation
* cuda arch and cuda 8 compatible

test=develop

test=develop

Add API.spec

test=develop

test=develop

test=develop

* flags

* "follow comment"

test=develop

scope's API modifies its internal state. And scope's
API can be called from multiple threads during traing.
Hence, we need locks to protect the scope's internal
states.

We can optimize it in the future. But the current
solution is buggy.

test=develop

* show detail error log on ci

* test

* fix memopt and dist

* update apispec

* will fix different batch issue test=develop

- add naive executor
- fix concurrency performance issue

test=develop

* add GraphNum

test=develop

* add graph number check in parallelExecutor

test=develop

* fix transformer_model bug

test=develop

* fix graph num

  extended test_text_classification ot use new op

test=develop

test=develop

- Added draft of new operator

- Added fused embedding fc lstm files

- First time embedding_fc_lstm_fuse_pass was invoked in
  test_text_classification

- Added Embedding pattern

- Not crashing

- Enabled draft of embedding_fc_lstm pass (does it job)

- First working (Seqcompute only) version

- Removed diagnostic comment

- First enabling of BatchCompute

- Disabling pass for embedding with is_sparse and is_distributed

- Cosmetics

- Style

- Style

* Add CUDA implementation for generate_proposals_op.
* Clean code.
* Update code.