* [Auto Parallel] Reorganize the fold structure

* [Auto Parallel] Fix some import errors

* [Auto Parallel] Speedup the completion process

* [Auto Parallel] Skip the property of dist_context when deepcopying

* [Auto Parallel] Remove the unnecessary print

* [Auto Parallel] Move some changes from 2.4 branch to develop

* Update engine.py

* [Auto Parallel] Fix a bug

* unique id for mesh

* rng ctrl

* support dropout

* register op

* adopt for recompute

* update unitest

* support pp

* update env setting

* update pass logic

* dist op support bf16

* backward cast update

* update setting

* update backward

* revert amp pass

* update fp16 backward logic

* register c_embedding bf16

* revert engine

* add unitest

* add unitest

* update unitest

* update cmake

* update math

* update math.py

* update unitest

* update unitest

* revise unitest

* revise unitest

* update unitest

* update unitest

* update unitest

* add FusedLinear pass

* add fused_op_list and renname PASSES to OP_FUSION

* add fused_passes_list to constants.py

* add test_passes.py

* fix test_fused_passes.py

* fix add if float(paddle.version.cuda()) >= 11.6:

* renamed test_fused_passes.py

* fix CMakeList.txt

* [Auto Parallel] Remove some fluid APIs

* [Auto Parallel] Fix the wrong import

* [Auto Parallel] Remove unnecessary comments

* [Auto Parallel] Fix the importing bug

* remove deps and prior comm

* grad comm fuse

* add deps for amp&global norm

* stage2 broadcast prior deps

* stage2 grad overlap

* stream_analyzer bugfix

* overlap enable

* dep op namescope

* depend support multiple inputs

* check finite deps

* stage2 param comm overlap

* Set kD2HStream

* grad comm hierarchical

* grad comm hierarchical

* new unitest
Co-authored-by: Nchenruibiao <chenruibiao@baidu.com>

* auto parallel bf16

* [AutoParallel] quantization pass support export

* support subgraph

* move_presist_var_to_global_block

* update unittest

* fix ci-coverage

* fix codestyle

* fix fake_dequantize_op

* remove unused var

* fix ci error and aprroval error

* add unittest for fp16 in test_dequant_linear

* replace mutable data

* fix unittest in non-cuda-core

* fix unittest
Co-authored-by: Ncarryyu <569782149@qq.com>
Co-authored-by: Nwufeisheng <wfs1997@163.com>

* isort all files

* revert conflicting files

* revert conflicting files

* revert conflicting files

* update config

* re-blacken python code

* temporarily disable date and diff_py_file

* skip a format

* [AutoParallel] add callbacks

* fix unittest

* fix dist_context

* fix engine

* fix cmakelist

* fix unittest's returns

* fix cmakelist

* [CodeStyle][F401] remove unused import in python/paddle/distributed

* remove pass

* empty commit

* Fix ValueError: list.remove(x): x not in list for meta_optimizer_names.

Fix ValueError: list.remove(x): x not in list for meta_optimizer_names.

* Fix split import.

Fix split import.

* add noqa after meta_optimizers in factory

* restort collective ops

* expand `import *`

* add noqa after required imports

* try to fix APIs without core.ops

* Revert "try to fix APIs without core.ops"

This reverts commit 6172beaf601e84bf61f2490c12c4739f0edaa5eb.

* fix an increment

* empty commit

* add noqa after required imports

* expand `import *`, fix ci error
Co-authored-by: NShuangchi He <34329208+Yulv-git@users.noreply.github.com>

* support input mask

* [Auto Parallel] Use c++ dist attr in the completion process

* [Auto Parallel] Add minor changes

* [Auto Parallel] Use c++ dist attr in the completion process

* [Auto Parallel] Add minor changes

* [Auto Parallel] Add the serialization process for dist attrs

* [Auto Parallel] Remove unnecessary comments

* [Auto Parallel] Fix some bugs

* [Auto Parallel] Fix the code style

* [Auto Parallel] Remove unnecessary impls

* [Auto Parallel] Fix the importing error

* [Auto Parallel] Fix the copy from bugs of op dist attr

* [Auto Parallel] Replace the use of constexpr if

* [Auto Parallel] Redesign the shard_tensor, shard_op and ProcessMesh

* [Auto Parallel] Change API of the completion unittest

* [Auto Parallel] Fix the bug when set_attr an int

* [Auto Parallel] Add the unittest for the serialization

* [Auto Parallel] Add some unit tests

* [Auto Paralle] Unify the strategy

* [Auto Parallel] Improve the engine api

* [Auto Parallel] Reset the changes made to the framework

* [Auto Parallel] Change the engine unittest

* [Auto Parallel] Update API of the completion and partitioner

* [Auto Parallel] Update unit tests using engine api

* update shard annotation

* [Auto Parallel] Remove the modifications of other modules

* [Auto Parallel] Add docs for APIs

* add new strategy

* [Auto Parallel] Replace the logger

* [Auto Parallel] Restore the test_program.py

* [Auto Parallel] Change the import rules

* [Auto Parallel] Add the examples for Engine

* [Auto Parallel] Do some minor changes

* [Auto Parallel] Remove yaml dependency

* [Auto Parallel] Fix the unittests

* add valid after train

* bug fix
Co-authored-by: Nzhaoyingli <zhaoyingli@baidu.com>
Co-authored-by: Ncaozhou <caozhou@radi.ac.cn>
Co-authored-by: Ncaozhou <48191911+Caozhou1995@users.noreply.github.com>

* adapt gradient merge

* fix op_role

* fix strategy

* adapt lazy init and fix pass

* add unittest

* update comment

* fix amp and sharding

* remove clip_by_norm

* add grad_clip pass

* add unittest

* add notes

* update func

* add dist_attr for new op

* add quant pass

* add clip_grad

* fix comments

* add unittest

* update logger

* add collate_fn

* fix number of inputs

* bugfix

* remove scaling

* support rescale_grad opt

* avoid sync with cpp in partition op

* delay eval & predict mode

* bugfix for gradient merge pass

* add 'to_static' in engine api

* fix cmakelist

* fixed bug for pass & engine

* fixed bug for benchmark GPT-3

* fix gradient merge

* bug fix

* update annotation

* use yapf to format all python file

* yapf exclude two unittests file for they rely on writing and reading file, and format will break them

* disable diff_py_file because too many diff files cause command following failed

* [Auto Parallel] Add the parallel tuner

* [Auto Parallel] Improve the parallel tuner and fix some bugs

* upodate cost model

* update import Resharder by dist op

* update cost model

* fix comp cost bug

* update cost model

* [Auto Parallel] Amend the dist attr for #processses=1

* update cost model and tuner

* update cost model and tuner

* update cost model and tuner

* update cluster

* update reshard

* [Auto Parallel] Add the estimation from the cost model

* [Auto Parallel] Reimplement the backup and restore functions

* [Auto Parallel] Fix the bugs of the parallel tuner

* [Auto Parallel] Update the engine api and dist context

* [Auto Parallel] Work around the high order grad problem

* [Auto Parallel] Add some miscellaneous improvements

* [Auto Parallel] Add a unittest for DistributedContext
Co-authored-by: Ncaozhou <caozhou@radi.ac.cn>

* use original id in dist_op_context.grad_op_id_to_op_id

* del assert

* remove redundant map

* slice data in dist_loader & flag to scale grad

* bug fix

* update unittest

* enable static

* [Auto Parallel] Refactor the engine api and parallelizer

* [Auto Parallel] Fix the default dist op for the slice op

* [Auto Parallel] Fix the format of planer.py

* [Auto Parallel] Fix a bug