LightningIRWandbLogger

class lightning_ir.main.LightningIRWandbLogger(name: str | None = None, save_dir: str | Path = '.', version: str | None = None, offline: bool = False, dir: str | Path | None = None, id: str | None = None, anonymous: bool | None = None, project: str | None = None, log_model: 'all' | bool = False, experiment: Run | RunDisabled | None = None, prefix: str = '', checkpoint_name: str | None = None, **kwargs: Any)[source]

Bases: WandbLogger

Lightning IR extension of the Weights & Biases Logger for enhanced experiment tracking.

This logger extends the PyTorch Lightning WandbLogger to provide improved file management and experiment tracking specifically tailored for Lightning IR experiments. It ensures that experiment files are properly saved in the WandB run’s files directory and handles the save directory management correctly.

Methods

Attributes

save_dir

Gets the save directory for experiment files and artifacts.

property save_dir: str | None

Gets the save directory for experiment files and artifacts.

This property returns the directory where WandB saves experiment files, logs, and artifacts. It handles the case where the experiment might not be properly initialized (DummyExperiment) and returns None in such cases to prevent errors.

Returns:

The absolute path to the WandB experiment directory where files

are saved, or None if the experiment is not properly initialized or WandB is running in offline/disabled mode.

Return type:

str | None