Merge pull request !1916 from lilei/add_proximal_ada_grad_optimizer

Merge pull request !1960 from liubuyu/master

Merge pull request !1957 from jiangjinsheng/issue_merge

Merge pull request !1933 from yoonlee666/edit-example

Merge pull request !1888 from YuJianfeng/adam

Merge pull request !1946 from lichen/remove_autoparallel_st

Merge pull request !1873 from yanghaoran/master

Merge pull request !1874 from h.farahat/string_tensor_updates

Merge pull request !1945 from vlne-v1/I1JXUP-resnet50-thor-assign-fail

- Support accepting Numpy of str as input
- Support batching strings
- Core logic of batch&pad is static
- Make Pad a utility function

Merge pull request !1947 from lianliguang/fix-bug-of-tbe-adapter

Merge pull request !1943 from zhaoting/master

Merge pull request !1940 from huanghui/single-batchnorm-fission-pass

Merge pull request !1936 from meixiaowei/master

This reverts commit 705c71a257.

Merge pull request !1911 from lianliguang/add-a-function-to-charge-the-node-input-or-output-if-is-a-scalar

Merge pull request !1931 from panfengfeng/fastrcnn_code_update

Merge pull request !1905 from baihuawei/cpulstm

Merge pull request !1101 from chentingting/gcn

Merge pull request !1927 from zongha/master

Merge pull request !1929 from laiyongqiang/mem

Merge pull request !1915 from VectorSL/bert-finetune

Merge pull request !1857 from yuchaojie/transformer

Merge pull request !1917 from jiangjinsheng/vm_FusedMUlAddL2loss

Merge pull request !1921 from liuxiao/ApplyProximalAdagrad-for-GE

Merge pull request !1906 from chenzhongming/master