当前位置：首页 > news >正文

深度学习-mmcv中build_runner实例化全流程详解

news 来源：原创 2025/5/21 5:52:31

1、实例化的例子

#运行的类型是IterBasedRunne
runner = dict(
type="IterBasedRunner",
max_iters=num_iters_per_epoch * num_epochs,
)
#实例化运行器，会返回一个IterBasedRunner的实例化对象
runner = build_runner(cfg.runner,default_args=dict(model=model,optimizer=optimizer,work_dir=cfg.work_dir,logger=logger,meta=meta,),
)

2、实例化的完整调用流程及关键注释
~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/builder.py：

#调用~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/utils/registry.py中的Registry类，实例化注册器，'runner'、'runner builder'对应__init__的形参name
RUNNERS = Registry('runner')
RUNNER_BUILDERS = Registry('runner builder')def build_runner_constructor(cfg: dict):#实例化~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/default_constructor.py中的DefaultRunnerConstructorreturn RUNNER_BUILDERS.build(cfg)def build_runner(cfg: dict, default_args: Optional[dict] = None):runner_cfg = copy.deepcopy(cfg)#参数配置文件中没提供'constructor'，这里采用默认运行构造器'DefaultRunnerConstructor'constructor_type = runner_cfg.pop('constructor','DefaultRunnerConstructor')runner_constructor = build_runner_constructor(dict(type=constructor_type,runner_cfg=runner_cfg,default_args=default_args))runner = runner_constructor()#调用构造器的__call__()方法return runner

~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/default_constructor.py

导入~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/builder.py中的RUNNER_BUILDERS, RUNNERS
from .builder import RUNNER_BUILDERS, RUNNERS@RUNNER_BUILDERS.register_module()
class DefaultRunnerConstructor:def __init__(self, runner_cfg: dict, default_args: Optional[dict] = None):if not isinstance(runner_cfg, dict):raise TypeError('runner_cfg should be a dict',f'but got {type(runner_cfg)}')self.runner_cfg = runner_cfgself.default_args = default_args#~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/builder.py中的runner = runner_constructor()会调用__call__因为self.runner_cfg中的type为IterBasedRunner，所以会实例化~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/iter_based_runner.py中的class IterBasedRunner(BaseRunner):def __call__(self):return RUNNERS.build(self.runner_cfg, default_args=self.default_args)

~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/iter_based_runner.py:

@RUNNERS.register_module()
#继承父类的__init__
class IterBasedRunner(BaseRunner):

~/anaconda3/envs/sparsedrive/lib/python3.7/site-packages/mmcv/runner/base_runner.py：

class BaseRunner(metaclass=ABCMeta):def __init__(self,model: torch.nn.Module,batch_processor: Optional[Callable] = None,optimizer: Union[Dict, torch.optim.Optimizer, None] = None,work_dir: Optional[str] = None,logger: Optional[logging.Logger] = None,meta: Optional[Dict] = None,max_iters: Optional[int] = None,max_epochs: Optional[int] = None) -> None:if batch_processor is not None:if not callable(batch_processor):raise TypeError('batch_processor must be callable, 'f'but got {type(batch_processor)}')warnings.warn('batch_processor is deprecated, please implement ''train_step() and val_step() in the model instead.',DeprecationWarning)# raise an error is `batch_processor` is not None and# `model.train_step()` exists.if is_module_wrapper(model):_model = model.moduleelse:_model = modelif hasattr(_model, 'train_step') or hasattr(_model, 'val_step'):raise RuntimeError('batch_processor and model.train_step()/model.val_step() ''cannot be both available.')else:assert hasattr(model, 'train_step')# check the type of `optimizer`if isinstance(optimizer, dict):for name, optim in optimizer.items():if not isinstance(optim, Optimizer):raise TypeError(f'optimizer must be a dict of torch.optim.Optimizers, 'f'but optimizer["{name}"] is a {type(optim)}')elif not isinstance(optimizer, Optimizer) and optimizer is not None:raise TypeError(f'optimizer must be a torch.optim.Optimizer object 'f'or dict or None, but got {type(optimizer)}')# check the type of `logger`if not isinstance(logger, logging.Logger):raise TypeError(f'logger must be a logging.Logger object, 'f'but got {type(logger)}')# check the type of `meta`if meta is not None and not isinstance(meta, dict):raise TypeError(f'meta must be a dict or None, but got {type(meta)}')self.model = modelself.batch_processor = batch_processorself.optimizer = optimizerself.logger = loggerself.meta = meta# create work_dirif isinstance(work_dir, str):self.work_dir: Optional[str] = osp.abspath(work_dir)mmcv.mkdir_or_exist(self.work_dir)elif work_dir is None:self.work_dir = Noneelse:raise TypeError('"work_dir" must be a str or None')# get model name from the model classif hasattr(self.model, 'module'):self._model_name = self.model.module.__class__.__name__else:self._model_name = self.model.__class__.__name__self._rank, self._world_size = get_dist_info()self.timestamp = get_time_str()self.mode: Optional[str] = Noneself._hooks: List[Hook] = []self._epoch = 0self._iter = 0self._inner_iter = 0if max_epochs is not None and max_iters is not None:raise ValueError('Only one of `max_epochs` or `max_iters` can be set.')self._max_epochs = max_epochsself._max_iters = max_iters# TODO: Redesign LogBuffer, it is not flexible and elegant enoughself.log_buffer = LogBuffer()