Merge pull request !7483 from 34bunny/GPU-ucs

Merge pull request !7548 from zhouyuanshen/master

Merge pull request !7106 from VectorSL/gpu-size_t

Merge pull request !6971 from chenweifeng/tensoradd_inplace

Merge pull request !6937 from baihuawei/lossmodify

Merge pull request !6818 from VectorSL/combine-mom

Merge pull request !6595 from baihuawei/ctc

Merge pull request !6574 from liubuyu/code_clean

Merge pull request !6584 from ZPaC/master-momentum-supports-use_nesterov

Merge pull request !6532 from VectorSL/bncast

Merge pull request !6321 from chenweifeng/codex

Merge pull request !6330 from VectorSL/pooling

Merge pull request !6317 from TuDouNi/relu6

Merge pull request !5790 from danishnxt/GPU_three

Merge pull request !5690 from JonathanY/rcnn

style fix

lint fixes

added check in NN layer for > 4 paddings, plus lint fix

fix python lint

lint fix

lint fix

updating to pytest asserts to improve testing

removed unnecc vars from test file fail checks

Merge pull request !4778 from Peilin/reshape-gpu-kernel-type-support

fix ci

Merge pull request !4706 from Peilin/smoothL1Loss-fix

Merge pull request !4475 from JonathanY/batchnormgrad

Merge pull request !4758 from VectorSL/nhwc-support

Merge pull request !4559 from chenweifeng/warning

fix pylint

changed doc and code for SmoothL1Loss to be same a dchip. fixed grad kernel

fix ci