WandbLogger

class WandbLogger(name: Optional[str] = None, entity: Optional[str] = None, project: Optional[str] = None, job_type: Optional[str] = None, group: Optional[str] = None, tags: Optional[List[str]] = None, plot: bool = True, table: bool = True, name_prefix: str = '', config: Optional[Dict[str, Any]] = None, log_model: bool = False)

Bases: etna.loggers.base.BaseLogger

Weights&Biases logger.

Create instance of WandbLogger.

Parameters

• name (Optional[str]) – Wandb run name.

• entity (Optional[str]) – An entity is a username or team name where you’re sending runs.

• project (Optional[str]) – The name of the project where you’re sending the new run

• job_type (Optional[str]) – Specify the type of run, which is useful when you’re grouping runs together into larger experiments using group.

• group (Optional[str]) – Specify a group to organize individual runs into a larger experiment.

• tags (Optional[List[str]]) – A list of strings, which will populate the list of tags on this run in the UI.

• plot (bool) – Indicator for making and sending plots.

• table (bool) – Indicator for making and sending tables.

• name_prefix (str) – Prefix for the name field.

• config (Optional[Dict[str, Any]]) – This sets wandb.config, a dictionary-like object for saving inputs to your job, like hyperparameters for a model or settings for a data preprocessing job.

• log_model (bool) –

Inherited-members

Methods

finish_experiment()	Finish experiment.
log(msg, **kwargs)	Log any event.
log_backtest_metrics(ts, metrics_df, ...)	Write metrics to logger.
log_backtest_run(metrics, forecast, test)	Backtest metrics from one fold to logger.
reinit_experiment()	Reinit experiment.
set_params(**params)	Return new object instance with modified parameters.
start_experiment([job_type, group])	Start experiment.
to_dict()	Collect all information about etna object in dict.

Attributes

experiment	Init experiment.
pl_logger	Pytorch lightning loggers.

finish_experiment()

Finish experiment.

log(msg: Union[str, Dict[str, Any]], **kwargs)

Log any event.

e.g. “Fitted segment segment_name” to stderr output.

Parameters

• msg (Union[str, Dict[str, Any]]) – Message or dict to log

• kwargs – Parameters for changing additional info in log message

Notes

We log dictionary to wandb only.

log_backtest_metrics(ts: TSDataset, metrics_df: pandas.core.frame.DataFrame, forecast_df: pandas.core.frame.DataFrame, fold_info_df: pandas.core.frame.DataFrame)

Write metrics to logger.

Parameters

• ts (TSDataset) – TSDataset to with backtest data

• metrics_df (pandas.core.frame.DataFrame) – Dataframe produced with etna.pipeline.Pipeline._get_backtest_metrics()

• forecast_df (pandas.core.frame.DataFrame) – Forecast from backtest

• fold_info_df (pandas.core.frame.DataFrame) – Fold information from backtest

log_backtest_run(metrics: pandas.core.frame.DataFrame, forecast: pandas.core.frame.DataFrame, test: pandas.core.frame.DataFrame)

Backtest metrics from one fold to logger.

Parameters

• metrics (pandas.core.frame.DataFrame) – Dataframe with metrics from backtest fold

• forecast (pandas.core.frame.DataFrame) – Dataframe with forecast

• test (pandas.core.frame.DataFrame) – Dataframe with ground truth

reinit_experiment()

Reinit experiment.

start_experiment(job_type: Optional[str] = None, group: Optional[str] = None, *args, **kwargs)

Start experiment.

Complete logger initialization or reinitialize it before the next experiment with the same name.

Parameters

• job_type (Optional[str]) – Specify the type of run, which is useful when you’re grouping runs together into larger experiments using group.

• group (Optional[str]) – Specify a group to organize individual runs into a larger experiment.

property experiment

Init experiment.

property pl_logger

Pytorch lightning loggers.