class lightning_fabric.loggers.TensorBoardLogger(root_dir, name='lightning_logs', version=None, default_hp_metric=True, prefix='', sub_dir=None, **kwargs)[source]

Bases: lightning_fabric.loggers.logger.Logger

Log to local file system in TensorBoard format.

Implemented using SummaryWriter. Logs are saved to os.path.join(root_dir, name, version). This is the recommended logger in Lightning Fabric.

  • root_dir (Union[str, Path]) – The root directory in which all your experiments with different names and versions will be stored.

  • name (Optional[str]) – Experiment name. Defaults to 'lightning_logs'. If it is the empty string then no per-experiment subdirectory is used.

  • version (Union[int, str, None]) – Experiment version. If version is not specified the logger inspects the save directory for existing versions, then automatically assigns the next available version. If it is a string then it is used as the run-specific subdirectory name, otherwise 'version_${version}' is used.

  • default_hp_metric (bool) – Enables a placeholder metric with key hp_metric when log_hyperparams is called without a metric (otherwise calls to log_hyperparams without a metric are ignored).

  • prefix (str) – A string to put at the beginning of all metric keys.

  • sub_dir (Union[str, Path, None]) – Sub-directory to group TensorBoard logs. If a sub_dir argument is passed then logs are saved in /root_dir/name/version/sub_dir/. Defaults to None in which case logs are saved in /root_dir/name/version/.

  • **kwargs (Any) – Additional arguments used by tensorboardX.SummaryWriter can be passed as keyword arguments in this logger. To automatically flush to disk, max_queue sets the size of the queue for pending logs before flushing. flush_secs determines how many seconds elapses before flushing.


from lightning.fabric.loggers import TensorBoardLogger

logger = TensorBoardLogger("path/to/logs/root", name="my_model")
logger.log_hyperparams({"epochs": 5, "optimizer": "Adam"})
logger.log_metrics({"acc": 0.75})

Do any processing that is necessary to finalize an experiment.


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

Return type:


log_graph(model, input_array=None)[source]

Record model graph.

  • model (Module) – the model with an implementation of forward.

  • input_array (Optional[Tensor]) – input passes to model.forward

Return type:


log_hyperparams(params, metrics=None)[source]

Record hyperparameters. TensorBoard logs with and without saved hyperparameters are incompatible, the hyperparameters are then not displayed in the TensorBoard. Please delete or move the previously saved logs to display the new ones with hyperparameters.

Return type:


log_metrics(metrics, step=None)[source]

Records metrics. This method logs metrics as soon as it received them.

  • metrics (Mapping[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:



Save log data.

Return type:


property experiment: tensorboardX.SummaryWriter

Actual tensorboard object. To use TensorBoard features anywhere in your code, do the following.


property log_dir: str

The directory for this run’s tensorboard checkpoint.

By default, it is named 'version_${self.version}' but it can be overridden by passing a string value for the constructor’s version parameter instead of None or an int.

property name: str

Get the name of the experiment.


The name of the experiment.

property root_dir: str

Gets the save directory where the TensorBoard experiments are saved.


The local path to the save directory where the TensorBoard experiments are saved.

property sub_dir: Optional[str]

Gets the sub directory where the TensorBoard experiments are saved.


The local path to the sub directory where the TensorBoard experiments are saved.

property version: Union[int, str]

Get the experiment version.


The experiment version if specified else the next version.

© Copyright Copyright (c) 2018-2023, Lightning AI et al...

Built with Sphinx using a theme provided by Read the Docs.