* [Fix] Make autocast compatible with mps

* Enhance unit test

* fix unit test

* clean the code

* fix unit test

* Update

* Ensure max_iters being integer number

* First commit

* add BaseTestTimeAugModel

* Add unit test

* move loop logic to test_step

* fix ddp test

* rename model to module

* optim __init__

* Fix as comment

* Fix as comment

* make val_step should not be called

* make tta do not inherit base model

* Fix unit test

* Enhance docstring

* Fix as comment

* Fix as comment

* minor refine

* minor refine

* minor refine

* fix unit test

* minor refine

* minor refine

* minor refine

* minor refine

* minor refine

* minor refine

* fix unit test

* fix loss mean

* update

* Upgrade the versions of pre-commit-hooks

* update zh-cn.yaml

* [Refactor] Refactor fileio but without breaking bc

* handle compatibility

* fix format

* modify io functions

* fix ut

* fix ut

* rename method names

* refine

* refine docstring

* fix ut in windows

* update ut

* minor fix

* ensure client is not None when closing it

* add more examples for list_dir_or_file interface

* refine docstring

* refine deprecated info

* fix ut

* add a description for lmdb docstring

* [Fix] Support compare numpy array dataset meta

* Add comment

* [Feature] Support convert BN to SyncBN by config

* make unit test compatible with cpu

* Fix as comment

* fix unit test

* change signature of convert_sync_batchnorm: rename sync_bn to implemention

* fix unit test

* fix unit test

* add set random seed fun

* fix conflict

* allign the previous version

* [Refactor]: modify interface of Visualizer.add_datasample (#365)

* [Refactor] Refactor data flow: refine `data_preprocessor`. (#359)

* refine data_preprocessor

* remove unused BATCH_DATA alias

* Fix type hints

* rename move_data to cast_data

* [Refactor] Refactor data flow: collate data in `collate_fn` of `DataLoader`  (#323)

* acollate data in dataloader

* fix docstring

* refine comment

* fix as comment

* refactor default collate and psedo collate

* foramt test file

* fix docstring

* fix as comment

* rename elem to data_item

* minor fix

* fix as comment

* [Refactor] Refactor data flow: `data_batch` argument of `Evaluator.process is a `dict` (#360)

* refine evaluator and metric

* compatible with new default collate

* replace default collate with pseudo

* Handle data_batch in metric

* fix unit test

* fix unit test

* fix unit test

* minor refine

* make data_batch optional

make data_batch optional

* rename outputs to predictions

* fix ut

* rename predictions to outputs

* fix docstring

* fix docstring

* fix unit test

* make outputs and data_batch to kwargs

* fix unit test

* keep signature of metric

* fix ut

* rename pred_sample arguments to data_sample(Visualizer)

* fix loop and ut

* [refactor]: Refactor model dataflow (#398)

* [Refactor] Refactor data flow: refine `data_preprocessor`. (#359)

* refine data_preprocessor

* remove unused BATCH_DATA alias

* Fix type hints

* rename move_data to cast_data

* refactor model data flow

tmp_commt

tmp commit

* make val_cfg and test_cfg optional

* roll back runner

* pass test mmdet

* fix as comment

fix as comment

fix ci in DataPreprocessor

* fix ut

* fix ut

* fix rebase main

* [Fix]: Fix test val ddp (#462)

* [Fix] Fix docstring and type hint of data flow (#463)

* Fix docstring of data flow

* change signature of hook

* fix unit test

* resolve conflicts

* fix lint

* Rename data to structure

* adjust the way to import module

* adjust the way to import module

* rename Structure to Data Structures in docs api

* rename structure to structures

* support using some modules of mmengine without torch

* fix circleci config

* fix circleci config

* fix registry ut

* minor fix

* move init method from model/utils to model/weight_init.py

* move init method from model/utils to model/weight_init.py

* move sync_bn to model

* move functions depending on torch to dl_utils

* format import

* fix logging ut

* add weight init in model/__init__.py

* move get_config and get_model to mmengine/hub

* move log_processor.py to mmengine/runner

* fix ut

* Add TimeCounter in dl_utils/__init__.py

* [Refactor] Refactor the import rule

* minor refinement

* add a comment

* Support build evaluator from list of built metric

* regist evaluator

* fix as comment

* add unit test

* support build optimizer wrapper from built Optimizer instance

* refine comments

* make default scope of runner be mmengine

* fix as comment

* [Enhancement] Add unit test for get_priority

* fix priority ut

* fix typo

Co-authored-by: Wenwei Zhang <40779233+ZwwWayne@users.noreply.github.com>

* fix wrong epoch andd iter when save bbest ckpt

* fix ut

* fix resume best ckpt unexpectedly

* minor refine

* fix unit test

* support training with data without metainfo

* clean the code

* clean the code

* [Enhance] Add `preprocess_cfg` as an argument of runner.

* Rename `preprocess_cfg` to `data_preprocessor`

* Fix docstring

* Support log enviroment information when initiate runner

* Fix unit test

* fix as comment, save world_size

* log gpu num

* clear code and reformat log

* minor refine

* fix as comment

* minor refine

* clean the code

* clean the code

* remove save world_size in meta

* add build function for scheduler

* add unit test

add unit test

* handle convert_to_iter in build_scheduler_from_cfg

* restore deleted code

* format import

* fix lint

* fix build multiple scheduler

* add new unit test

* fix comment and error message

* fix comment and error message

* extract _parse_scheduler_cfg

* always call build_param_scheduler during train and resume. If there is only one optimizer, the defaut value for sheduler will be a list, otherwise there is multiple optimizer, the default value of sheduler will be a dict

* minor refine

* rename runner test exp name

* fix as comment

* minor refine

* fix ut

* only check parameter scheduler

* minor refine

* make scheduler default to None

* fix bc breaking

* refine warning message

* fix as comment

* fix as comment

* fix lint

* Fix resume message hub and save metainfo in messagehub

* fix as comment

* [Fix]: Fix resume bug

* [Fix]: Change last_checkpoint check logic

* [Fix]: Fix lint

* [Fix]: Change warning to print_log

* [Fix] Fix weight initializing and registry logging.

* sync params

* resolve comments

* [Enhance] Auto set the `end` of param schedulers.

* Add log output and unit test

* Update docstring

* Update unit tests of `CosineAnnealingParamScheduler`.

* fix resume message_hub

* add unit test

* support resume from messagehub

* minor refine

* add comment

* fix typo

* update docstring

* 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 amp

* fix amp

* make logic consistent with torch.autocast

* support multiple device

* fix as comment

* fix as comment

* avoid circle import

* fix build multiple runner error

* fix comments

* fix cpu ci

* 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 unit test of autocast

* fix compatiblity of unit test of optimizerwrapper

* clean code

* fix as comment

* fix docstring