* [NPU] add run_program_op_npu

* add run_program_op_npu ut

* optimizer conv alog speed

* code polish

* remove useless code

* fix compile error

* fix cpu compile error

* not use cudnn alog t

* add search cache max number

* polish code

* fix cache test bug

* add groups data format to conv args

* fix cache test bug

* fix cudnn_deterministic bug

* fix test switch auto tune bug

* fix test swith autotune bug;

* fix conv cache bug

* fix cache test error

* fix cache test bug

* fix windows mac compile error

* fix workspace search error

* update cudnn cache

* fix cache test bug; test=develop

* fix autotune swith test error

* polish code

* oplish code

* support high order differential with data parallel overlap

* update unitest

* add python final state

* fix bug

* fix bugs

* fix bug

* fix bug

* revert impl, final state mul not support selected rows

* fix softmax use cudnn error

* add softlable false unitest

* revert loss.py

* fix utest

* fix utest

* fix utest

* fix log

* fix random utest

* [DOC FIX]fix code demo of auc

* [doc fix] fix doc of auc

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* Adapt minlength attr for bincount

* fixed1

* fix 2

* fixre

* test complext var

* delete logic no needed

* fix to_tensor_static

* code style

* del

* while support for python container.
It is convenient to convert more dynamic graph codes into static graphs.

* cond support python container

* add quant pass

* bugfix

* remove scaling

* support rescale_grad opt

* add unitest

* new executor support program._graph

* fix program._program if program is Program()

* move distribute_fpn_proposals

* fix some code

* fix yaml bugs

* add set dtype

* move proposal_impl to funcs

* fix compile bugs

* [CustomDevice] add profiler apis

* migrate CalculateEstOccupancy into cuda_tracer

* update

* add ut

* Add int8 support for matmul+elementwiae_add fuse

* Corrections after review and ernie test fix

* merge conv_concat_relu to conv_act

* fix typo

* extend unit test

* reuse existing gpd

* codestyle

* enforce mkldnn conv

some python c api use final state

* optimize python api to speed up eager exec

* optimize python api to speed up eager exec

* optimize python api to speed up eager exec

* pylayer detach output tensor if it is equal with input

* pylayer detach output tensor if it is equal with input

* add max_p without test

* add test of max_p

* make max_p consistent with paddle.maximum

* add forward inplace final state api

* fix bug

* fix reshape

* fix coverage

* add inplace info for erfinv, lerp, put_along_axis

* fix put_along_axis infer_meta

* fix format

* update yaml

* fix

* fix layernormTrt meanVar alloc bug

* Fix random op depenency and lr_shedule bugs for standalone executor

* Fix CI errors

* Fix CI errors

* Fix CI errors

* add group_norm dyanmic plugin

* polish default param of XXX_interp_test, the same default value with XXX_interp_np

* set default value data_layout=NCHW, cause C++ end treat NCDHW the same way as NCHW

* add merged_momentum *test=kunlun

* add merged_momentum *test=kunlun

* add fp16 to merged_momentum,*test=kunlun

* change dist_model.cc

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun