* [Eager] Support eager grad interface, draft version

* Support eager grad interface with allow_unused and multi startup_op

* Fix code format

* Fix allow_unused case, return PyNone if tensor not initialize

* Support output's stop_gradient related to create_graph

* Support grad exception case in eager mode, fix coverage CI

* Update ToPyObject, return PyNone if not initialize

* AccumulationNode add FLAGS_retain_grad_for_all_tensor

* Fix ci issue

* Fix CI issue

* fix, use core.eager.Tensor

* Add func SetBufferSlotRankZeros for GradTensorHolder

* Support retain_graph by using ClearTensorWrappers

* Support retain_graph by using ClearTensorWrappers

* Update retain_graph and no_grad_vars related test case

* Update code gen logic for ClearTensorWrappers

* Fix by override statement

* fix override func args

* Support retain_graph, update unit tests

* Updated ClearTensorWrappers logic

* fix grad python interface

* Use deep copy and update unit tests

* Polish code

* Polish code

* Fix CI issue, Deep copy only use when user set grad_tensors

* Fix CI, use Backward instead RunBackward

* Fix CI, Declare kernel explicitly in test file

* Polish, remove vector of TensorWrapper

* Refactor the logic of grad/backward, polish codes

* Update code after merge upstream develop

* Polish after merge upstream develop

* Update to adapt new GradNodeBase superclass

* Fix error introduced during conflict resolution

* Update purify potential_startup_nodes logic

* Fix errors

* Polish code

* Remove useless args for ToPyObject

* Remove useless TensorWrappersSet

* Fix code-format, re-install pre-commit

* Fix pre-process logic for potential_startup_ops

* Update unit tests, use eager mode

* fix copy_ problem by doing it with phi copy

* improve test coverage

* refactor copy with sr kernel

* Move the computation of moving average scale to device

* Use register to save local maximum in a thread

* init commit

* correct namespace

* Add yaml config for pool2d

* Fix CI error

* Fix code format error

* [Phi] Migrate mode_op and mode_grad_op into Phi

* fix omp

* add ifdef

* migrate infershape

* modify according reviewer

* Refactor elementwise op grad classes

* Add more refactor changes

* Revert set layout and format deletion

* Fix failing elementwise test

* Quantize elementwise mul op

* Parametrize elementwise functions

* Fix code formatting

* move roi_align_grad kernel

* move roi_align grad kernel and infershape to phi

* remove roi_align infershape

* [PHI] Migrate roll op

* 【phi】migrate eigh op to phi (#40213)

* migrate eigh to phi

* optimize code

* modify code according to comment

* conflict resolution

* [PHI] Migrate roll op

* [PHI] Fix converage of roll_sig

* [PHI] Fix infermate of roll_sig

* [Phi] Fix unittest coverage of roll op

* [PHI] Fix infermeta in unary

* [PHI] Fix parameter type of roll op

* [PHI] Fix parameter type of roll op

* [PHI] Fix parameter of roll op
Co-authored-by: Ncrystal <62974595+Zjq9409@users.noreply.github.com>

* [PHI] Migrate index_select op

* [PHI] Fix bug in test_variable

* [PHI] migrate index_select op

* [kp]fix bug that cannot fallback to CPU normally in XPU KP

* fix bug in static graph

* migrate multiplex op kernel

* migrate qr cpu kernel

* migrate tril_triu op kernel

* fix multiplex kernel

* add kernel sig

* fix dependence and bug

* fix multiplex error

* fix npu include error

* fix conflict

* fix conflict and delete tril_triu

* fix date and multiplex input

* adapt header file order

* fix header file include

* fix conflict

* delete cholesky_solve_op.h

* delete triangular_solve_op.h

* add grid sample phi kernel

* add grid sample phi kernel and remove original kernel

* replace mutable_data by alloc

* [Auto Parallel] Support the auto completion of while_op

* [Auto Parallel] Improve the completion algorithms

* [Auto Parallel] Fix bugs for ernie inference

* [Auto Parallel] Remove attrs which cannot be pickled

* [Auto Parallel] make the dims_mappings of LodTensorArray vars empty

* [Auto Parallel] Fix bugs for the ernie inference in the pipeline parallel

* [Auto Parallel] Remove unncessary comments

* [Auto Parallel] Fix a bug of the CMakeLists

* [Auto Parallel] Use the newest APIs to write the unit test

* [Auto Parallel] Remove unnecessary statements

* add determinant phi kernel

* remove original determinant op kernel

* add determinant grad [hi kernel

* fix determinant test failed

* remove original determinant grad op kernel

* [phi] move matrix_power op

* MatrixInverse fluid -> phi

* modify the CMake to fix compile bug

* delete useless comment

* mutable memory -> phi Alloc

* modify the include file

* modify the include file

* fix bug in CI compiler

* [phi]modify the shape OP and move inferMeta of shape,matrix_pow,multi_dot

* delete useless comment

* fix bug in CI

* modify after review

* add expert count op

add ut for expert_count

* update UT only for cuda

* fix for rocm

* update ut

* add moe module

* add expert count op

add ut for expert_count

* update UT only for cuda

* update ut

* add moe module

* make expert count private

* rename expert count op
Co-authored-by: Nhlygit66666 <2570058140@qq.com>

* run python api in eager model and filter the out in argument list

* fix code

* Fixed issues with generated scale operator

* Fixed minor issues

* [NPU] add AMP O1 support

* [NPU] fix NOTE and warnings

* add phi gather kernel

* update year

* remove original gather opkernel

* add gather grad phi kernels

* remove origin gather grad kernel

* fix failed npu and xpu

* fix xpu compile failed

* - Prototype of third solution

- fix

- compilation fixes

- fix

- fixe

- fix

- fix

- compilation fix

- comment fix

- lint

update mkldnn conv_elementwise_add_fuse_pass ut

- NHWC changes to prelu

- alhpa dims

- UT fix

- fix to UT

- lint

- Some fixes

- added to BWD of prelu NHWC support

- reverted removal of resetting cu_layout in clearing of caching

* - Small changes

* - compilation fix

* - fix

* - fix

* lint

* - fixes after internal review

* - compilation fix

* - lint

* shard_id

* format

[phi] Transfer lgamma, kldiv_loss, isclose, cumprod kernels into phi and pass the tests of these four kernels (#39770)

* tranfer and pass the lgamma unittest

* merge and pass the test

* transfer kldiv_loss and kldiv_loss_grad; pass the unitest

* trafer the isclose and cumprod kernel

* change PT_REGISTER -> PD_REGISTER

* fix by code review

* fix by code review

* fix

* remove enforce include dependence from scalar

* fix

* fix by code review

* fix by code review

* add reduce_min kernel

* remove raw reduce_min kernel

* add reduce min

* add reduce any all impl

* add bool reduce Kernel

* remove raw any/all kernel

* add any all kernel

* rm comment

* Added more signposts to dygraph profiling

* Fixed minor issues

* Refactored signpost names

* Fixed typo

* Removed debug codes

* Fixed typo

* Adjusted signpost names

* Fixed issues from branch merge

* move one hot to phi; test=develop

* fix bugs; test=develop

* fix bugs; test=develop

* add infer meta; test=develop

* fix bugs; test=develop

* resolve confilct

* resolve confilct

* fix bug;

* fix error; test=develop

* update; test=develop

* polish code; test=develop

* add one api in eager mode; test=develop

* add one hot test; test=develop

* remove use less code; test=develop

* fix bug; test=develop

* polish code; test=develop

* polish code; test=develop