Pop the value of the key. If class attribute ``lazy`` is True, the LazyObject will be built and returned. Args: key (str)

Yield the keys and values of the dictionary. If class attribute ``lazy`` is False, the value of ``LazyObject`` or ``LazyAttr`` will b

Get the registry record. If `key`` represents the whole object name with its module information, for example, `mmengine.model.BaseMod

Overrides this method to call :meth:`BaseDataPreprocessor.to` additionally. Returns: nn.Module: The model itself.

Build a runner from config. Args: cfg (ConfigType): A config used for building runner. Keys of ``cfg`` can see :m

Convert a version string into a tuple of integers. This method is usually used for comparing two versions. For pre-release versions: alpha <

Launch training. Returns: nn.Module: The model after training.

Print a log message. Args: msg (str): The message to be logged. logger (Logger or str, optional): If the type of logger is

Get subclass instance by name if the name exists. If corresponding name instance has not been created, ``get_instance`` will create a

Return the attribute of the imported object. Returns: Any: attribute of the imported object.

Register a module. A record will be added to ``self._module_dict``, whose key is the class name or the specified name, and value is t

Returns a dictionary containing log scalars, runtime information and resumed keys, which should be resumed. The returned ``state_dict

Get the value of the key. If class attribute ``lazy`` is True, the LazyObject will be built and returned. Args: key (str)

Returns: list: Contains all keys in data_fields.

Get runtime information by key. If the key does not exist, this method will return default information. Args: key (str):

Build a Config instance from config file. Args: filename (str or Path): Name of config file. use_predefined_variables

Check whether a file path is a file. Args: filepath (str or Path): Path to be checked whether it is a file. Returns:

Override this method to make sure the LazyObject will not be built during updating.

Call the step method of optimizer.

Yield the values of the dictionary. If class attribute ``lazy`` is False, the value of ``LazyObject`` or ``LazyAttr`` will be built a

Check whether a file path exists. Args: filepath (str or Path): Path to be checked whether exists. Returns:

Get latest created ``MMLogger`` instance. :obj:`MMLogger` can call :meth:`get_current_instance` before any instance has been created,

Asserts that ``actual`` and ``expected`` are close. A wrapper function of ``torch.testing.assert_allclose``. Args: actual (Any): Actu

Build a module from config dict when it is a class configuration, or call a function from config dict when it is a function configuration. If

Overrides this method to call :meth:`BaseDataPreprocessor.cpu` additionally. Returns: nn.Module: The model itself.

Recursively delete a directory tree. Args: dir_path (str or Path): A directory to be removed. Examples: >>>

Call all hooks. Args: fn_name (str): The function name in each hook to be called, such as "before_train_epoch".

Scan a directory to find the interested directories or files in arbitrary order. Note: :meth:`list_dir_or_file` returns t

Update runtime information. The key corresponding runtime information will be overwritten each time calling ``update_info``.

Returns the total aggregated statistic across all operators for the requested module. Args: module_name (str): The submod

Build an instance. Build an instance by calling :attr:`build_func`. Args: cfg (dict): Config dict needs to be built.

Overrides this method to call :meth:`BaseDataPreprocessor.cuda` additionally. Returns: nn.Module: The model itself.

Check whether the object has a method. Args: method (str): The method name to check. obj (object): The object to check. Retu

Override the `logging.Formatter.format`` method `. Output the message according to the specified log level. Args: record

r"""Concatenate all file paths. Join one or more filepath components intelligently. The return value is the concatenation of filepath

Scan a directory to find the interested directories or files in arbitrary order. Note: Petrel has no concept of directori

Returns a copy of the :class:`JitModelAnalysis` object, keeping all settings, but on a new model or new inputs. Args: new

Launch test. Returns: dict: A dict of metrics on testing set.

Deep copy the current data element. Returns: BaseDataElement: The copy of current data element.

Dump config to file or return config text. Args: file (str or Path, optional): If not specified, then the object is d

Transform file to variables dictionary. Args: filename (str): Name of config file. use_predefined_variables (bool, op

Return a file backend based on the prefix of uri or backend_args. Args: uri (str or Path): Uri to be parsed that contains the file prefix

Read data from a given ``filepath`` with 'rb' mode. Note: There are two types of return values for ``get``, one is ``bytes``

Get ``HistoryBuffer`` instance by key. Note: Considering the large memory footprint of history buffers in the post-tr

Read data from a given ``filepath`` with 'r' mode. Args: filepath (str or Path): Path to read data. encoding (str): T

Finish all operations, related to checkpoint. This function will get the appropriate file client, and the directory to save these che

Get annotation by index and automatically call ``full_init`` if the dataset has not been fully initialized. Args: idx (in

A wrapper of ``torch.autocast`` and ``toch.cuda.amp.autocast``. Pytorch 1.5.0 provide ``torch.cuda.amp.autocast`` for running in mixed precis

Get priority value. Args: priority (int or str or :obj:`Priority`): Priority. Returns: int: The priority value.

Check whether it is a sequence of some type. Args: seq (Sequence): The sequence to be checked. expected_type (type or tuple): Exp

Update :attr:_log_scalars. Update ``HistoryBuffer`` in :attr:`_log_scalars`. If corresponding key ``HistoryBuffer`` has been created,

Return the recently updated values in log histories. Returns: np.ndarray: Recently updated values in log histories.

Given a model and an input to the model, compute the per-operator Gflops of the given model. Adopted from https://github.com/facebookrese

Returns: iterator: An iterator object whose element is (key, value) tuple pairs for ``data``.

Detach all tensors in data.

Return the mean of the latest ``window_size`` values in log histories. If ``window_size is None`` or ``window_size > len(self._log_hi

Set or change key-value pairs in ``data_field`` by parameter ``data``. Args: data (dict): A dict contains annotations of

Map ``filepath`` to a string path whose prefix will be replaced by :attr:`self.path_mapping`. Args: filepath (str or Path

Convert all data in the config to a builtin ``dict``. Args: keep_imported (bool): Whether to keep the imported field.

Sets if warnings for unsupported operators are shown. Defaults to True. Counts of unsupported operators may be obtained from :meth:`u

Convert a ``filepath`` to standard format of petrel oss. If the ``filepath`` is concatenated by ``os.path.join``, in a Windows enviro

Build an iter-based instance of this scheduler from an epoch-based config.

Build optimizer wrapper. If ``optim_wrapper`` is a config dict for only one optimizer, the keys must contain ``optimizer``, and ``typ

Draw featmap. - If `overlaid_image` is not None, the final output image will be the weighted sum of img and featmap. - If

Check whether a file path is a file. Args: filepath (str or Path): Path to be checked whether it is a file. Returns:

Register a backend. Args: name (str): The name of the registered backend. backend (class, optional): The backend class to be regi

Save checkpoint to file. Args: checkpoint (dict): Module whose params are to be saved. filename (str): Checkpoint filename.

Copy a file src to dst and return the destination file. src and dst should have the same prefix. If dst specifies a directory, the fi

Check whether a file path exists. Args: filepath (str or Path): Path to be checked whether exists. Returns:

Get vis backend by name. Args: name (str): The name of vis backend Returns: BaseVisBackend: The vis backend

Check if a module is a model wrapper. The following 4 model in MMEngine (and their subclasses) are regarded as model wrappers: DataParallel,

Loads log scalars, runtime information and resumed keys from ``state_dict`` or ``message_hub``. If ``state_dict`` is a dictionary ret

Write data to a given ``filepath`` with 'w' mode. Note: ``put_text`` should create a directory if the directory of ``

Remove a file. Args: filepath (str, Path): Path to be removed.

Launch validation. Returns: dict: A dict of metrics on validation set.

Save the checkpoint and synchronize buffers after each evaluation epoch. Args: runner (Runner): The runner of the trainin

Draw single or multiple text boxes. Args: texts (Union[str, List[str]]): Texts to draw. positions (Union[np.ndarray,

Get the shapes from a jit value object. Args: val (torch._C.Value): jit value object. Returns: list(int): return a list of i

Check whether it is a list of some type. A partial method of :func:`is_seq_of`.

Returns the currently existing device type. Returns: str: cuda | npu | mlu | mps | musa | cpu.

Returns True if cuda devices exist.

Load data from json/yaml/pickle files. This method provides a unified api for loading data from serialized files. ``load`` supports loading

Record the scalar data to tensorboard. Args: name (str): The scalar identifier. value (int, float, torch.Tensor, np.n

Build evaluator. Examples of ``evaluator``:: # evaluator could be a built Evaluator instance evaluator = Evaluator(m

Build parameter schedulers. ``build_param_scheduler`` should be called after ``build_optim_wrapper`` because the building logic will

Check whether a file path exists. Args: filepath (str or Path): Path to be checked whether exists. Returns: