WandBLogger

class torchtune.utils.metric_logging.WandBLogger(project: str, entity: Optional[str] = None, group: Optional[str] = None, **kwargs)

Logger for use w/ Weights and Biases application (https://wandb.ai/). For more information about arguments expected by WandB, see https://docs.wandb.ai/ref/python/init.

Parameters:

• project (str) – WandB project name

• entity (Optional[str]) – WandB entity name

• group (Optional[str]) – WandB group name

• **kwargs – additional arguments to pass to wandb.init

Example

>>> from torchtune.utils.metric_logging import WandBLogger
>>> logger = WandBLogger(project="my_project", entity="my_entity", group="my_group")
>>> logger.log("my_metric", 1.0, 1)
>>> logger.log_dict({"my_metric": 1.0}, 1)
>>> logger.close()

Raises:

ImportError – If wandb package is not installed.

Note

This logger requires the wandb package to be installed. You can install it with pip install wandb. In order to use the logger, you need to login to your WandB account. You can do this by running wandb login in your terminal.

close() → None

Close log resource, flushing if necessary. Logs should not be written after close is called.

log(name: str, data: Union[Tensor, ndarray, int, float], step: int) → None

Log scalar data.

Parameters:

• name (str) – tag name used to group scalars

• data (Scalar) – scalar data to log

• step (int) – step value to record

log_dict(payload: Mapping[str, Union[Tensor, ndarray, int, float]], step: int) → None

Log multiple scalar values.

Parameters:

• payload (Mapping[str, Scalar]) – dictionary of tag name and scalar value

• step (int) – step value to record