* fix lars optitmizer bug

* Update optimizer.py

* fix doc for enable api

* test=document_fix

* [Auto Parallel] Support the auto completion of while_op
* align infer  accuracy

* add cast_grad phi kernel

* refie unittest

* refien unittest

* refine unittest

* refine include header path

* refien xpu cast unittest

* refine code

* support multi_dims for tril_triu, *test=kunlun

* support multi_dims for tril_triu, *test=kunlun

* support multi_dims for tril_triu, *test=kunlun

* update xpu.cmake date, support multi_dims for tril_triu, *test=kunlun

* Update transforms.py

* Update transforms.py

* Update transforms.py

* Update functional.py

* refactor eager flags

* fix flags error when we switch from eager to dygraph

* fix ci problem

* fix ci

* fix ci

* merge develop and fix code style

* merge develop and fix code style

* fix op test error

* fix op test error

* fix op test error

* fix op test error

* fix op test error

* merge develop

* 1. add the python api grad 2. add final and intermediate state vlog 3. change the python_api error logic

* add python api or close the check_eager=True

* fix the compatibility

* matmul

* disable unittests: test_elementwise_add_op test_scatter_nd_op test_gather_nd_op test_scatter_op test_index_sample_op test_elementwise_add_mkldnn_op

* refine the logic of prepara_parameter logic

* fix Tensor(gpu) 2 Scalar segment fault.

* add multi-attribute. (test_unsqueeze_op); add python_sig_out for customizing op sig out

* fix some bugs, support python_out_sig

* [NPU] add merged_momentum

* fix

* fix device

* Scalar support marking data_type in yaml

* fix code-gene bug

* support get_item where the index is a bool scalar tensor

* add unittests for supporting get_item where the index is a bool scalar tensor

* support intermediate for saprse api

* close intermediate in yaml

* fix dygraph_api dep for eager

* approve amp for intermediate_dygraph

* add amp_utils for intermediate_dygraph

* add amp needcast check for mlu & npu

* test unittest

* add SetGradNode for set_stop_gradient && add checktensor for GradientHooks

* refine code

* refien unittest of imperative_amp for new dygraph

* inplace api skip amp

* add test_imperative_qat_amp for intermediate amp

* refine code

* refine test_amp ci strategy

* refine unittest code

* refine amp_utils code

* refine amp getpromotetype for some special op

* refine unittest code

* # This is a combination of 10 commits.
# The first commit's message is:
add expert count op

add ut for expert_count

# This is the 2nd commit message:

update UT only for cuda

# This is the 3rd commit message:

fix for rocm

# This is the 4th commit message:

update ut

# This is the 5th commit message:

add moe module

# This is the 6th commit message:

add expert count op

add ut for expert_count

# This is the 7th commit message:

update UT only for cuda

# This is the 8th commit message:

update ut

# This is the 9th commit message:

add moe module

# This is the 10th commit message:

make expert count private

* add assign pos op

* fix upper num name

* add api _assign pos

* add ut for assign pos op

* update date

* fix for win

* update for test (timeout)

* fix ut

* update

* fix ut for number count
Co-authored-by: Nhlygit66666 <2570058140@qq.com>

* test=document_fix , fix launch doc

* test=document_fix , fix typo

* Fix rnn, wmt16 docs;test=document_fix

* Fix wmt14 docs;test=document_fix

* Add more description;test=document_fix

* [Auto Parallel] gradient merge pass support dist attribute

Polish optest: refine the optest parameter logic. support name, dtype, out, output in arbitrary position (#40824)

* 1. add the python api grad 2. add final and intermediate state vlog 3. change the python_api error logic

* add python api or close the check_eager=True

* fix the compatibility

* matmul

* disable unittests: test_elementwise_add_op test_scatter_nd_op test_gather_nd_op test_scatter_op test_index_sample_op test_elementwise_add_mkldnn_op

* refine the logic of prepara_parameter logic

* fix Tensor(gpu) 2 Scalar segment fault.

* Refine eager run_program OP for dy2st UT

* append run_program error string and refine run_program_grad

* remove some comments

* refine ConstructXGradTensors

* refactor cost model

* added missing BF16 activations

* added softplus bf16

* minor change

* disabled tests for GPU

* [NPU] add npu support for conv3d and conv3d_grad

* [NPU] delete failed unittests due to Ascend not support

* [NPU] delete debug codes

* [NPU] optimize codes, notest

* [NPU] remove const_cast

* [NPU] optimize for remove const_cast

* [NPU] fix written errors

* fix benchmark and communicator config

* fix bugs of the_one_ps

* multi program and fix bug in optimizer

* multi program in the_one_ps

* public commcontext

* ps optimizer multi programs

* cvm & datanorm backend

* fix dim

* fix unittest

* fix

* the one ps merge

* remove comm

* add DownpourLiteWorker

* all

* fix

* fix

* device worker downpour lite

* fix

* fix bug in global shuffle

* save inference model

* fix & add log

* fix

* remove log

* fix

* fix save summary

* fix

* fix pscore

* fix

* fix

* fix

* fix

* fix

* remove logs

* fix

* fix

* fix

* fix

* fix

* add some comments

* fix
Co-authored-by: Nesythan <esythan@126.com>

* add dist_saver and update engine

* add dist_saver and update engine

* disable scatter case in test_inplace_eager_fluid

* Update register_hook logic

* Add register_hook test cases
Co-authored-by: Npangyoki <pangyoki@126.com>

* suppor sharding api

* support multi api for sharding in eager

* support multi api for sharding in eager

* fix test

* fix test coverage

* fix error; test=develop

* update

* close some yaml

* fix backward attrite error; test=develop

* add div test

* polish code; test=develop

* remove none gbk charactor;

* remove some yaml;

* fix optional bug

* recover yaml config

* resolve confilct; test=develop

* close div; test=develop

* fix some slice bug, test=develop

* eager slice, test=develop

* eager slice, test=develop

* refine, test=develop

* refine, test=develop

* fix bug, test=develop

* refine, test=develop

* rename function name, test=develop