setup_tb_logging

ignite.handlers.logger_utils.setup_tb_logging(output_path, trainer, optimizers=None, evaluators=None, log_every_iters=100, *, trainer_metric_names='all', evaluator_metric_names='all', **kwargs)

Method to setup TensorBoard logging on trainer and a list of evaluators. Logged metrics are:

• Training metrics, e.g. running average loss values

• Learning rate(s)

• Evaluation metrics

Parameters:

• output_path (str) – logging directory path

• trainer (Engine) – trainer engine

• optimizers (Optimizer | dict[str, torch.optim.optimizer.Optimizer] | None) – single or dictionary of torch optimizers. If a dictionary, keys are used as tags arguments for logging.

• evaluators (Engine | dict[str, ignite.engine.engine.Engine] | None) – single or dictionary of evaluators. If a dictionary, keys are used as tags arguments for logging.

• log_every_iters (int) – interval for loggers attached to iteration events. To log every iteration, value can be set to 1 or None.

• trainer_metric_names (str | list[str]) – list of trainer metric names to plot or a string “all” to plot all available metrics.

• evaluator_metric_names (str | list[str]) – list of evaluator metric names to plot or a string “all” to plot all available metrics.

• kwargs (Any) – optional keyword args to be passed to construct the logger.

Return type:

TensorboardLogger

Changed in version 0.6.0: Added trainer_metric_names and evaluator_metric_names parameters.

Returns:

TensorboardLogger

Parameters:

• output_path (str) –

• trainer (Engine) –

• optimizers (Optimizer | dict[str, torch.optim.optimizer.Optimizer] | None) –

• evaluators (Engine | dict[str, ignite.engine.engine.Engine] | None) –

• log_every_iters (int) –

• trainer_metric_names (str | list[str]) –

• evaluator_metric_names (str | list[str]) –

• kwargs (Any) –

Return type:

TensorboardLogger

Examples

from ignite.handlers.logger_utils import setup_tb_logging
# Assume `trainer`, `evaluator`, and `optimizer` are already defined
tb_logger = setup_tb_logging(
    output_path="experiments/tb_logs",
    trainer=trainer,
    optimizers=optimizer,
    evaluators={"validation": evaluator},
    log_every_iters=100
)

# Logger instance can be closed
tb_logger.close()