pytorch_lightning.loggers.base module

class pytorch_lightning.loggers.base.DummyExperiment

Bases: object

Dummy experiment

nop(**kw)

class pytorch_lightning.loggers.base.DummyLogger

Bases: pytorch_lightning.loggers.base.LightningLoggerBase

Dummy logger for internal use. Is usefull if we want to disable users logger for a feature, but still secure that users code can run

log_hyperparams(params)

Record hyperparameters.

Parameters

params – Namespace containing the hyperparameters

log_metrics(metrics, step)

Records metrics. This method logs metrics as as soon as it received them. If you want to aggregate metrics for one specific step, use the agg_and_log_metrics() method.

Parameters

• metrics – Dictionary with metric names as keys and measured quantities as values

• step – Step number at which the metrics should be recorded

property experiment

Return the experiment object associated with this logger.

property name

Return the experiment name.

property version

Return the experiment version.

class pytorch_lightning.loggers.base.LightningLoggerBase(agg_key_funcs=None, agg_default_func=numpy.mean)

Bases: abc.ABC

Base class for experiment loggers.

Parameters

• agg_key_funcs (Optional[Mapping[str, Callable[[Sequence[float]], float]]]) – Dictionary which maps a metric name to a function, which will aggregate the metric values for the same steps.

• agg_default_func (Callable[[Sequence[float]], float]) – Default function to aggregate metric values. If some metric name is not presented in the agg_key_funcs dictionary, then the agg_default_func will be used for aggregation.

Note

The agg_key_funcs and agg_default_func arguments are used only when one logs metrics with the agg_and_log_metrics() method.

_aggregate_metrics(metrics, step=None)

Aggregates metrics.

Parameters

• metrics (Dict[str, float]) – Dictionary with metric names as keys and measured quantities as values

• step (Optional[int]) – Step number at which the metrics should be recorded

Return type

Tuple[int, Optional[Dict[str, float]]]

Returns

Step and aggregated metrics. The return value could be None. In such case, metrics are added to the aggregation list, but not aggregated yet.

static _convert_params(params)

Return type

Dict[str, Any]

_finalize_agg_metrics()

This shall be called before save/close.

static _flatten_dict(params, delimiter='/')

Flatten hierarchical dict, e.g. {'a': {'b': 'c'}} -> {'a/b': 'c'}.

Parameters

• params (Dict[str, Any]) – Dictionary containing the hyperparameters

• delimiter (str) – Delimiter to express the hierarchy. Defaults to '/'.

Return type

Dict[str, Any]

Returns

Flattened dict.

Examples

>>> LightningLoggerBase._flatten_dict({'a': {'b': 'c'}})
{'a/b': 'c'}
>>> LightningLoggerBase._flatten_dict({'a': {'b': 123}})
{'a/b': 123}

_reduce_agg_metrics()

Aggregate accumulated metrics.

static _sanitize_params(params)

Returns params with non-primitvies converted to strings for logging.

>>> params = {"float": 0.3,
...           "int": 1,
...           "string": "abc",
...           "bool": True,
...           "list": [1, 2, 3],
...           "namespace": Namespace(foo=3),
...           "layer": torch.nn.BatchNorm1d}
>>> import pprint
>>> pprint.pprint(LightningLoggerBase._sanitize_params(params))  
{'bool': True,
 'float': 0.3,
 'int': 1,
 'layer': "<class 'torch.nn.modules.batchnorm.BatchNorm1d'>",
 'list': '[1, 2, 3]',
 'namespace': 'Namespace(foo=3)',
 'string': 'abc'}

Return type

Dict[str, Any]

agg_and_log_metrics(metrics, step=None)

Aggregates and records metrics. This method doesn’t log the passed metrics instantaneously, but instead it aggregates them and logs only if metrics are ready to be logged.

Parameters

• metrics (Dict[str, float]) – Dictionary with metric names as keys and measured quantities as values

• step (Optional[int]) – Step number at which the metrics should be recorded

close()

Do any cleanup that is necessary to close an experiment.

Return type

None

finalize(status)

Do any processing that is necessary to finalize an experiment.

Parameters

status (str) – Status that the experiment finished with (e.g. success, failed, aborted)

Return type

None

abstract log_hyperparams(params)

Record hyperparameters.

Parameters

params (Namespace) – Namespace containing the hyperparameters

abstract log_metrics(metrics, step=None)

Records metrics. This method logs metrics as as soon as it received them. If you want to aggregate metrics for one specific step, use the agg_and_log_metrics() method.

Parameters

• metrics (Dict[str, float]) – Dictionary with metric names as keys and measured quantities as values

• step (Optional[int]) – Step number at which the metrics should be recorded

save()

Save log data.

Return type

None

update_agg_funcs(agg_key_funcs=None, agg_default_func=numpy.mean)

Update aggregation methods.

Parameters

• agg_key_funcs (Optional[Mapping[str, Callable[[Sequence[float]], float]]]) – Dictionary which maps a metric name to a function, which will aggregate the metric values for the same steps.

• agg_default_func (Callable[[Sequence[float]], float]) – Default function to aggregate metric values. If some metric name is not presented in the agg_key_funcs dictionary, then the agg_default_func will be used for aggregation.

abstract property experiment

Return the experiment object associated with this logger.

Return type

Any

abstract property name

Return the experiment name.

Return type

str

property save_dir

Return the root directory where experiment logs get saved, or None if the logger does not save data locally.

Return type

Optional[str]

abstract property version

Return the experiment version.

Return type

Union[int, str]

class pytorch_lightning.loggers.base.LoggerCollection(logger_iterable)

Bases: pytorch_lightning.loggers.base.LightningLoggerBase

The LoggerCollection class is used to iterate all logging actions over the given logger_iterable.

Parameters

logger_iterable (Iterable[LightningLoggerBase]) – An iterable collection of loggers

close()

Do any cleanup that is necessary to close an experiment.

Return type

None

finalize(status)

Do any processing that is necessary to finalize an experiment.

Parameters

status (str) – Status that the experiment finished with (e.g. success, failed, aborted)

Return type

None

log_hyperparams(params)

Record hyperparameters.

Parameters

params (Union[Dict[str, Any], Namespace]) – Namespace containing the hyperparameters

Return type

None

log_metrics(metrics, step=None)

Records metrics. This method logs metrics as as soon as it received them. If you want to aggregate metrics for one specific step, use the agg_and_log_metrics() method.

Parameters

• metrics (Dict[str, float]) – Dictionary with metric names as keys and measured quantities as values

• step (Optional[int]) – Step number at which the metrics should be recorded

Return type

None

save()

Save log data.

Return type

None

property experiment

Return the experiment object associated with this logger.

Return type

List[Any]

property name

Return the experiment name.

Return type

str

property version

Return the experiment version.

Return type

str

pytorch_lightning.loggers.base.merge_dicts(dicts, agg_key_funcs=None, default_func=numpy.mean)

Merge a sequence with dictionaries into one dictionary by aggregating the same keys with some given function.

Parameters

• dicts (Sequence[Mapping]) – Sequence of dictionaries to be merged.

• agg_key_funcs (Optional[Mapping[str, Callable[[Sequence[float]], float]]]) – Mapping from key name to function. This function will aggregate a list of values, obtained from the same key of all dictionaries. If some key has no specified aggregation function, the default one will be used. Default is: None (all keys will be aggregated by the default function).

• default_func (Callable[[Sequence[float]], float]) – Default function to aggregate keys, which are not presented in the agg_key_funcs map.

Return type

Dict

Returns

Dictionary with merged values.

Examples

>>> import pprint
>>> d1 = {'a': 1.7, 'b': 2.0, 'c': 1, 'd': {'d1': 1, 'd3': 3}}
>>> d2 = {'a': 1.1, 'b': 2.2, 'v': 1, 'd': {'d1': 2, 'd2': 3}}
>>> d3 = {'a': 1.1, 'v': 2.3, 'd': {'d3': 3, 'd4': {'d5': 1}}}
>>> dflt_func = min
>>> agg_funcs = {'a': np.mean, 'v': max, 'd': {'d1': sum}}
>>> pprint.pprint(merge_dicts([d1, d2, d3], agg_funcs, dflt_func))
{'a': 1.3,
 'b': 2.0,
 'c': 1,
 'd': {'d1': 3, 'd2': 3, 'd3': 3, 'd4': {'d5': 1}},
 'v': 2.3}

pytorch_lightning.loggers.base.rank_zero_experiment(fn)

Returns the real experiment on rank 0 and otherwise the DummyExperiment.

Return type

Callable