* fix resume message_hub

* add unit test

* support resume from messagehub

* minor refine

* add comment

* fix typo

* update docstring

* first commit

* Support modify base config and add unit test

* remove import mmengine in config

* add unit test

* fix lint

* add unit test

* move RemoveAssignFromAST to config utils

* git add utils

* fix format issue in test file

* refine unit test

* refine unit test

* clean code

* fix as comment

* fix as comment

* add get_registry_by_scope method

* add unit test and docstring example

* rename get_registry_by_scope to switch_scope_and_registry

* move build function to registry/builder

* fix docstring

* rename builder->registry_builder, move build_from_cfg to registry_builder

rename builder->registry_builder, move build_from_cfg to registry_builder

* rename registry_builder to build_function

rename registry_builder to build_function

* fix docstring and type hint

* rename build_function to build_functions

* fix abs criterion error

* update

* fix missing device ids in wrap_model

* clean the code

* use default broadcast_buffers

* refine MMSeparateDistributedDataParallel

* rename tmp variable

* refine docstring

* add type hints

* refactor docstring of ddp model

* add  arg in docstring

* minor refine

* better ddp link

* fix cbds

* fix

* fix optimizer wrapper counts

* fix ut

* fix amp

* fix amp

* make logic consistent with torch.autocast

* support multiple device

* fix as comment

* fix as comment

* avoid circle import

* [Enhance] Support scheduling betas with MomentumScheduler.

* enhance ut

* test adam betas

* enhance ut

* enhance ut

* fix build multiple runner error

* fix comments

* fix cpu ci

* Fix ema hook and add unit test

* save state_dict of ema.module

save state_dict of ema.module

* replace warning.warn with MMLogger.warn

* fix as comment

* fix bug

* fix bug

* support dynamic interval in iterbasedtrainloop

* update typehint

* update typehint

* add dynamic interval in epochbasedtrainloop

* update

* fix

Co-authored-by: luochunhua.vendor <luochunhua@pjlab.org.cn>

* fix bug of not save-best in iteration-based training

* revise the unit test

* fix unit test of autocast

* fix compatiblity of unit test of optimizerwrapper

* clean code

* fix as comment

* fix docstring

* assert image shape before forward

* add unit test

* enhance error message

* allow gray image input

* fix as comment

* fix unit test

* fix unit test

* impl lr and momentum visualizer

* provide fakerun

* refine data_root and data_prefix params

* modify unittest

* [Feature]: Set different seed for diff rank

* [Feature]: Add log

* [Fix]: Fix lint

* [Fix]: Fix docstring

* [Fix]: Fix sampler seed

* [Fix]: Fix log bug

* [Fix]: Change diff_seed to diff_rank_seed

* [Fix]: Fix lint

* [Feat] Support mps

* fix docstring

* Fix docstring format of BaseDataElement

* fix docstring

* fix RuntimeError of SyncBuffersHook

* add UT

* [Feat] Support revert syncbn

* use logger.info but not warning

* fix info string

* add autocast wrapper

* fix docstring

* fix docstring

* fix compare version

* fix unit test

* fix incompatible arguments

* fix as comment

* fix unit test

* rename auto_cast to autocast

* [Feat] Support save best ckpt

* reformat code

* rename function and reformat code

* fix logging info

* LogProcessor support custom significant digit

* rename to num_digits

* modify cuda() to to()

* rollback load_checkpoint

* refine runner

* add TODO

* [Refactor]:Refactor `after_val_epoch` to make it output metric by epoch

* add an option for user to choose the way of outputing metric

* rename variable

* reformat docstring

* add type alias

* reformat code

* add test function

* add comment and test code

* add comment and test code

* [Feat] Support FSDP Training

* fix version comparison

* change param format and move `FSDP_WRAP_POLICY` to wrapper file

* add docstring and type hint,reformat code

* fix type hint

* fix typo, reformat code