* fix bug

* fix bug

* fix bug

* fix bug

* fix bug

* Add virtual pp and dp overlap

* add sharding/dp overlap

* add dp/vpp overlap

* fix code

* fix log

* make sharding reduce mode by default

* Update dygraph_sharding_optimizer.py

* Update hybrid_parallel_optimizer.py

* Update pipeline_parallel.py

* new_frl_shard_redece

* add mp guard

* add test

* add p2p calc stream

* rm code

* rm code

* rm assert

* rm code

* polish

* polish

* polish

* polish

* polish

* polish

* opt memory

* rm args

* add rm ptr

* add timer to pp (#53831)

* [Hybrid Performance] Sharding stage 1 PP/VP overlap (#54312)

[CodeStyle][C416][C417] rewrite unnecessary comprehension with function call and use generator instead of map (#52140)

* codestyle c416 c417

* fix error

* fix inc

* unify all C4 rules into one

* fix inc

---------
Co-authored-by: NSigureMo <sigure.qaq@gmail.com>

* isort all files

* revert conflicting files

* revert conflicting files

* revert conflicting files

* [CodeStyle][py2][U008] remove unnecessary args in `super()`

* remove remained args

* revert changes in test_pylayer_op

* Revert "revert changes in test_pylayer_op"

This reverts commit ff185a9ae738afac3b0264f61bde6c6b7f72e7c4.

* revert some changes in example code

* add unbalanced data

* fix utest

* update config

* re-blacken python code

* temporarily disable date and diff_py_file

* skip a format

* Fix bug of reduce_sum op. When input.numel() > INT32_MAX, its result
is wrong.

* Support allow_partial switch, which can be configure in
pipeline_configs. If sent tensor are not the same from
different hosts, they shouldn't been sent partially and
then concated as a whole tensor.

* Change name allow_partial to enable_partial_send_recv.

* Add global variable _enable_partial_send_recv

* [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>

* 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

* fix utest

* fix time

* [HybridParallel]Support fp16 in dygraph hybrid parallel

* update

* update

* update for recompute

* add unittest of pp+fp16

* add unittest of recompute+fp16

* update

* modify ut

* add no_sync for parameters sync

* add pipeline for moe

* support grad group

* fix single card condition

* sharding support dp

* sharding support mp

* sharding support pp

* add cache for send_recv

* add eval_batch for pipeline

* add eval batch for pipelineparallel

* add style code