* fix or backend many inputs bug

* fix or backend many inputs bug

* fix or backend many inputs bug

* fix or backend many inputs bug

* code format

* code format

* cpplint fix 3

* cpplint fix 3

* cpplint fix 3

* cpplint fix 3

* cpplint fix 2

* cpplint fix 2

resort .cu headers, set clang-format not sort include block and consider .cu as main source file (#43633)

* consolidate conv act passes

* generalize conv_activation

* integrate conv+act tests

* code style format

* whitespaces

* remove timeout from old tests

* implement comments from review

* restore ut

* whitespace

* code style

* transpose

* fixes after review

* method for gettin act

* Change Paddle_enforce error type

* code format

* add missing opcompats

* [MLU] add mlu kernel for elementwise_max_grad

* [MLU] modify mlu kernel elementwise_min_grad impl

* add argmin and argsort ops (#800)

* add argmin and arsort ops

* Add dot bmm ops (#803)

* add bmm

* add dot op

* clean CreateConst

* clean CreateCast

* add activation ops (#808)

* add activation ops

* fix 1function-redefined error

* fix for layer_norm

* minor fix

This reverts commit 1ec626b1.

* support more dimensions

* fix

* add new API/OP(csr->csr) of SparseTensor softmax

* fix comment

* gpups default config and dataset

* codestyle

* add unittest

* code style

* add dymf to gpups

* codestyle

* add static.nn.cvm import

* PSERVER_DEBUG

* add fs config to worker desc

* update unittest

* unittest

* remove gpups unittest

* remove gpups unittest

* static check

* fix to_sparse_csr

* Support optional residual add in fused_attention and fused_feedforward.

* Add checkpoint and add the check of add_residual when pre_layer_norm is false.

* Add TODO and change the python api to add add_residual argument.