d3rlpy.logging.TensorboardAdapterFactory

class d3rlpy.logging.TensorboardAdapterFactory(root_dir='tensorboard_logs')[source]

TensorboardAdapterFactory class.

This class instantiates TensorboardAdapter object.

Parameters:

root_dir (str) – Top-level log directory.

Methods

create(algo, experiment_name, n_steps_per_epoch)[source]

Creates LoggerAdapter.

This method instantiates LoggerAdapter with a given experiment_name. This method is usually called at the beginning of training.

Parameters:

  • algo (AlgProtocol) – Algorithm.

  • experiment_name (str) – Experiment name.

  • steps_per_epoch – Number of steps per epoch.

  • n_steps_per_epoch (int) –

Return type:

TensorboardAdapter