MessageHub — mmengine 0.10.7 documentation (original) (raw)

class mmengine.logging.MessageHub(name, log_scalars=None, runtime_info=None, resumed_keys=None)[source]¶

Message hub for component interaction. MessageHub is created and accessed in the same way as ManagerMixin.

MessageHub will record log information and runtime information. The log information refers to the learning rate, loss, etc. of the model during training phase, which will be stored as HistoryBuffer. The runtime information refers to the iter times, meta information of runner etc., which will be overwritten by next update.

Parameters:

• name (str) – Name of message hub used to get corresponding instance globally.
• log_scalars (dict, optional) – Each key-value pair in the dictionary is the name of the log information such as “loss”, “lr”, “metric” and their corresponding values. The type of value must be HistoryBuffer. Defaults to None.
• runtime_info (dict, optional) – Each key-value pair in the dictionary is the name of the runtime information and their corresponding values. Defaults to None.
• resumed_keys (dict, optional) – Each key-value pair in the dictionary decides whether the key in _log_scalars and_runtime_info will be serialized.

Note

Key in _resumed_keys belongs to _log_scalars or_runtime_info. The corresponding value cannot be set repeatedly.

Examples

create empty MessageHub.

message_hub1 = MessageHub('name') log_scalars = dict(loss=HistoryBuffer()) runtime_info = dict(task='task') resumed_keys = dict(loss=True)

create MessageHub from data.

message_hub2 = MessageHub( name='name', log_scalars=log_scalars, runtime_info=runtime_info, resumed_keys=resumed_keys)

classmethod get_current_instance()[source]¶

Get latest created MessageHub instance.

MessageHub can call get_current_instance() before any instance has been created, and return a message hub with the instance name “mmengine”.

Returns:

Empty MessageHub instance.

Return type:

MessageHub

get_info(key, default=None)[source]¶

Get runtime information by key. If the key does not exist, this method will return default information.

Parameters:

• key (str) – Key of runtime information.
• default (Any , optional) – The default returned value for the given key.

Returns:

A copy of corresponding runtime information if the key exists.

Return type:

Any

get_scalar(key)[source]¶

Get HistoryBuffer instance by key.

Note

Considering the large memory footprint of history buffers in the post-training, get_scalar() will not return a reference of history buffer rather than a copy.

Parameters:

key (str) – Key of HistoryBuffer.

Returns:

Corresponding HistoryBuffer instance if the key exists.

Return type:

HistoryBuffer

load_state_dict(state_dict)[source]¶

Loads log scalars, runtime information and resumed keys fromstate_dict or message_hub.

If state_dict is a dictionary returned by state_dict(), it will only make copies of data which should be resumed from the sourcemessage_hub.

If state_dict is a message_hub instance, it will make copies of all data from the source message_hub. We suggest to load data fromdict rather than a MessageHub instance.

Parameters:

state_dict (dict or MessageHub) – A dictionary contains keylog_scalars runtime_info and resumed_keys, or a MessageHub instance.

Return type:

None

property log_scalars_: OrderedDict_¶

Get all HistoryBuffer instances.

Note

Considering the large memory footprint of history buffers in the post-training, get_scalar() will return a reference of history buffer rather than a copy.

Returns:

All HistoryBuffer instances.

Return type:

OrderedDict

pop_info(key, default=None)[source]¶

Remove runtime information by key. If the key does not exist, this method will return the default value.

Parameters:

• key (str) – Key of runtime information.
• default (Any , optional) – The default returned value for the given key.

Returns:

The runtime information if the key exists.

Return type:

Any

property runtime_info_: OrderedDict_¶

Get all runtime information.

Returns:

A copy of all runtime information.

Return type:

OrderedDict

state_dict()[source]¶

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

The returned state_dict can be loaded by load_state_dict().

Returns:

A dictionary contains log_scalars, runtime_info andresumed_keys.

Return type:

dict

update_info(key, value, resumed=True)[source]¶

Update runtime information.

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

Note

The resumed argument needs to be consistent for the samekey.

Examples

message_hub = MessageHub(name='name') message_hub.update_info('iter', 100)

Parameters:

• key (str) – Key of runtime information.
• value (Any) – Value of runtime information.
• resumed (bool) – Whether the corresponding HistoryBuffercould be resumed.

Return type:

None

update_info_dict(info_dict, resumed=True)[source]¶

Update runtime information with dictionary.

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

Note

The resumed argument needs to be consistent for the sameinfo_dict.

Examples

message_hub = MessageHub(name='name') message_hub.update_info({'iter': 100})

Parameters:

• info_dict (str) – Runtime information dictionary.
• resumed (bool) – Whether the corresponding HistoryBuffercould be resumed.

Return type:

None

update_scalar(key, value, count=1, resumed=True)[source]¶

Update :attr:_log_scalars.

Update HistoryBuffer in _log_scalars. If corresponding keyHistoryBuffer has been created, value and count is the argument of HistoryBuffer.update, Otherwise, update_scalarwill create an HistoryBuffer with value and count via the constructor of HistoryBuffer.

Examples

message_hub = MessageHub(name='name')

create loss HistoryBuffer with value=1, count=1

message_hub.update_scalar('loss', 1)

update loss HistoryBuffer with value

message_hub.update_scalar('loss', 3) message_hub.update_scalar('loss', 3, resumed=False) AssertionError: loss used to be true, but got false now. resumed keys cannot be modified repeatedly'

Note

The resumed argument needs to be consistent for the samekey.

Parameters:

• key (str) – Key of HistoryBuffer.
• value (torch.Tensor or np.ndarray or int or float) – Value of log.
• count (torch.Tensor or np.ndarray or int or float) – Accumulation times of log, defaults to 1. count will be used in smooth statistics.
• resumed (str) – Whether the corresponding HistoryBuffercould be resumed. Defaults to True.

Return type:

None

update_scalars(log_dict, resumed=True)[source]¶

Update _log_scalars with a dict.

update_scalars iterates through each pair of log_dict key-value, and calls update_scalar. If type of value is dict, the value should be dict(value=xxx) or dict(value=xxx, count=xxx). Item inlog_dict has the same resume option.

Note

The resumed argument needs to be consistent for the samelog_dict.

Parameters:

• log_dict (str) – Used for batch updating _log_scalars.
• resumed (bool) – Whether all HistoryBuffer referred in log_dict should be resumed. Defaults to True.

Return type:

None

Examples

message_hub = MessageHub.get_instance('mmengine') log_dict = dict(a=1, b=2, c=3) message_hub.update_scalars(log_dict)

The default count of a, b and c is 1.

log_dict = dict(a=1, b=2, c=dict(value=1, count=2)) message_hub.update_scalars(log_dict)

The count of c is 2.